JP6683752B2 - Non-invasive determination of fetal or tumor methylome by plasma - Google Patents

Description

FIELD The present disclosure relates generally to the determination of methylation patterns (methylome) of DNA, and more particularly to the detection of a mixture of DNAs from different genomes (eg, fetal and maternal, or tumor and normal cell, respectively). It relates to analyzing a biological sample containing (eg plasma) to determine the methylation pattern (methylome) of a small number of genomes. The use of the determined methylome is also described.

CROSS REFERENCE TO RELATED APPLICATIONS This application is incorporated by reference in US Provisional Patent Application No. 61 / 830,571, "Tumor Detection In Plasma Using Measurement Status And Copy Number", filed June 3, 2013, filed September 20, 2012. U.S. Provisional Patent Application No. 61 / 703,512 "Method of Determining The Whore Genome DNA Methylation Status of The Placenta by Massively Application for 15 years in the 15th month of non-profitable May 13th of the month of 13th of May. Application No. 13 / 842,209 of "Non-Invasive Determination" f a Methylome Of Fetus Or Tumor From Plasma ", a PCT application claiming priority to, in their entirety are hereby incorporated by reference for all purposes.

Background Embryonic and fetal development is a complex process that involves a series of highly regulated genetic and epigenetic phenomena. Cancer development is also a complex process that typically involves multiple genetic and epigenetic processes. Abnormalities in the epigenetic control of developmental processes are associated with infertility, spontaneous abortion, intrauterine developmental abnormalities, and postnatal course. DNA methylation is one of the most studied epigenetic mechanisms. Methylation of DNA most commonly occurs in the context of adding a methyl group to the 5'carbon of cytosine residues between CpG dinucleotides. Cytosine methylation adds a layer of control to gene transcription and DNA function. For example, hypermethylation of CpG dinucleotide-rich gene promoters, called CpG islands, is usually associated with repression of gene function.

  Despite the importance of epigenetic mechanisms in coordinating developmental processes, human embryonic and fetal tissues are not readily available for analysis (and tumors may be unavailable as well) . Studies on the dynamic changes of these epigenetic processes in human prenatal health and disease are virtually impossible. Extraembryonic tissues, especially the placenta, have become one of the major tools for such studies, as they can be obtained as part of the prenatal diagnostic procedure or after delivery. However, such tissue requires an invasive procedure.

  The DNA methylation properties of human placenta have been of interest to researchers for decades. The human placenta exhibits numerous unique physiological features associated with DNA methylation. At a global level, placental tissue is hypomethylated compared to most somatic tissues. At the genetic level, the methylation status of a given genomic locus is a unique feature unique to placental tissue. Both global and locus-specific methylation properties show gestation-dependent changes. Imprinted genes, ie genes whose expression is dependent on the parental origin of the allele, play important functions in the placenta. The placenta has been described as pseudo-malignant and hypermethylation of several tumor suppressor genes has been observed.

  Studies of the DNA methylation properties of placental tissues have provided insights into the pathophysiology of pregnancy-related or development-related diseases such as preeclampsia and intrauterine growth restriction. Disorders in genomic imprinting are said to be associated with developmental disorders such as Prader-Willi syndrome and Angelman syndrome. Variable characteristics of genomic imprinting and global DNA methylation in placenta and fetal tissues have been observed in pregnancy caused by assisted reproductive medicine (H Hiura et al. 2012 Hum Reprod; 27: 2541-2548). Several environmental factors, such as mother's smoking (KE Harborth et al. 2013 Epigenomics; 5: 37-49), mother's dietary factor (X Jiang et al. 2012 FASEB J; 26: 3563-3574), and diabetes. Mother's metabolic status (N Hajj et al., Diabetes. Doi: 10.2337 / db12-0289) has been implicated in epigenetic abnormalities in offspring.

Despite efforts SUMMARY decades to study the Mechiromu fetal or tumors and, in order to monitor the dynamic changes throughout the course of the disease, such as pregnant or malignant tumors, what is available practical means There was no. Therefore, it is desired to provide a method for noninvasively analyzing all or part of a fetal methylome and a tumor methylome.

  Embodiments provide systems, methods, and devices for determining and using the methylation properties of various tissues and samples. Examples are provided. Methylation profile is estimated for fetal / tumor tissue based on comparison of plasma methylation (or other samples with cell-free DNA such as urine, saliva, genital effluent) to maternal / patient methylation profiles can do. If the sample has a mixture of DNAs, tissue-specific alleles can be used to identify DNA from the fetus / tumor and determine the methylation profile for that fetal / tumor tissue. The methylation profile can be used to determine copy number variation in the fetal / tumor genome. Methylation markers for the fetus have been identified by various techniques. The methylation profile can be determined by determining the size parameter of the size distribution of the DNA fragments and a reference value for the size parameter can be used to determine the methylation level.

  In addition, methylation levels can be used to determine cancer levels. In the cancer context, detecting cancer by measuring changes in methylome in plasma (eg, for screening purposes), monitoring (eg, detecting response following anti-cancer treatment, and detecting cancer recurrence). ), And prognosis (eg, for measuring the burden of cancer cells in the body, or for staging purposes, or for estimating the likelihood of death from a disease or disease progression or metastatic process) Becomes

  A better understanding of the nature and advantages of the embodiments of the present invention may be gained by reference to the following detailed description and the accompanying drawings.

10 shows the results of Table 100 of maternal blood, placenta, and maternal plasma sequencing according to embodiments of the present invention. 10 shows the results of Table 100 of maternal blood, placenta, and maternal plasma sequencing according to embodiments of the present invention. 3 shows the methylation density within a 1 Mb window of a sample sequenced according to an embodiment of the invention. Shown are plots of β values against the methylation index: (A) maternal blood cells, (B) chorionic villous specimen, (C) term placenta tissue. Shown are plots of β values against the methylation index: (A) maternal blood cells, (B) chorionic villous specimen, (C) term placenta tissue. Shown are plots of β values against the methylation index: (A) maternal blood cells, (B) chorionic villous specimen, (C) term placenta tissue. Figure 3 shows a bar graph of the percentage of methylated CpG sites in plasma and blood cells collected from adult males and non-pregnant adult females: (A) autosomal chromosome, (B) X chromosome. Figure 3 shows a bar graph of the percentage of methylated CpG sites in plasma and blood cells collected from adult males and non-pregnant adult females: (A) autosomal chromosome, (B) X chromosome. Shows plots of methylation densities of corresponding loci in blood cell DNA and plasma DNA: (A) non-pregnant adult female, (B) adult male. Shows plots of methylation densities of corresponding loci in blood cell DNA and plasma DNA: (A) non-pregnant adult female, (B) adult male. Figure 4 shows a bar graph of the percentage of methylated CpG sites between samples taken from pregnant women: (A) autosomal chromosome, (B) X chromosome. Figure 4 shows a bar graph of the percentage of methylated CpG sites between samples taken from pregnant women: (A) autosomal chromosome, (B) X chromosome. 3 shows a bar graph of methylation levels of different repeat classes of the human genome in maternal blood, placenta and maternal plasma. 1 shows a Circos plot 700 of a first trimester sample. 3 shows a Circos plot 750 of a third trimester sample. Figure 4 shows a comparative plot of methylation densities of genomic tissue DNA versus maternal plasma DNA for CpG sites surrounding the informative single nucleotide polymorphism. Figure 4 shows a comparative plot of methylation densities of genomic tissue DNA versus maternal plasma DNA for CpG sites surrounding the informative single nucleotide polymorphism. Figure 4 shows a comparative plot of methylation densities of genomic tissue DNA versus maternal plasma DNA for CpG sites surrounding the informative single nucleotide polymorphism. Figure 4 shows a comparative plot of methylation densities of genomic tissue DNA versus maternal plasma DNA for CpG sites surrounding the informative single nucleotide polymorphism. 6 is a flowchart illustrating a method 900 for determining a first methylation property from a biological sample of an organism according to an embodiment of the invention. 6 is a flow chart illustrating a method 1000 of determining a first methylation characteristic from a biological sample of an organism according to an embodiment of the invention. 6 shows a graph of the execution of a prediction algorithm using maternal plasma data and fetal DNA concentration fractions according to embodiments of the invention. 6 shows a graph of the execution of a prediction algorithm using maternal plasma data and fetal DNA concentration fractions according to embodiments of the invention. 1 is a table 1200 detailing 15 selected genomic loci for methylation prediction, according to an embodiment of the invention. 11 shows a graph 1250 showing the estimated categories of 15 selected genomic loci and their corresponding methylation levels in placenta. 6 is a flowchart of a method 1300 for detecting fetal chromosomal abnormalities from a biological sample of a female subject who has been pregnant with at least one fetus. 6 is a flow chart of a method 1400 for identifying methylation markers by comparing placental methylation profile to maternal methylation profile according to embodiments of the present invention. FIG. 13 is a table 1500 showing the performance of the DMR identification algorithm using first trimester data with reference to 33 previously reported trimester markers. FIG. 15 is a table 1550 showing the performance of the DMR identification algorithm compared to placental samples obtained at parturition using the third trimester data. 1B is a table 1600 showing the number of loci predicted to be hypermethylated or hypomethylated based on direct analysis of maternal plasma bisulfite sequencing data. 1D is a plot 1700 showing the size distribution of maternal plasma DNA, non-pregnant female control plasma DNA, placental DNA and peripheral blood DNA. 1750 is a plot 1750 of size distribution and methylation profile of maternal plasma, adult female control plasma, placental tissue and adult female control blood. 3 is a plot of methylation density and size of plasma DNA molecules, according to an embodiment of the invention. 3 is a plot of methylation density and size of plasma DNA molecules, according to an embodiment of the invention. FIG. 3B shows a plot 1900 of methylation density and size of sequenced reads for adult non-pregnant women. 1950, showing size distribution and methylation characteristics of fetal-specific and maternal-specific DNA molecules in maternal plasma. 3 is a flow chart of a method 2000 for estimating the methylation level of DNA in a biological sample of an organism, according to an embodiment of the invention. FIG. 2B is a table 2100 showing pre-operative plasma and tissue sample methylation densities of hepatocellular carcinoma (HCC) patients. FIG. 2B is a table 2150 showing the number of sequence reads and the sequencing depth achieved for each sample. Table 220, showing that the autosomal methylation densities ranged from 71.2% to 72.5% in healthy control plasma samples. 3 shows the methylation densities of buffy coat, tumor tissue, non-neoplastic liver tissue, pre-operative plasma and post-operative plasma of HCC patients. 3 shows the methylation densities of buffy coat, tumor tissue, non-neoplastic liver tissue, pre-operative plasma and post-operative plasma of HCC patients. FIG. 2B is a plot 2400 showing the methylation density of pre-operative plasma obtained from HCC patients. FIG. 3B is a plot 2450 showing the methylation density of post-operative plasma obtained from HCC patients. The Z values of plasma DNA methylation densities of plasma samples of HCC patients before surgery (plot 2500) and after surgery (plot 2550) were used, using plasma methylome data of 4 healthy control patients as a reference to chromosome 1. Show. The Z values of plasma DNA methylation densities of plasma samples of HCC patients before surgery (plot 2500) and after surgery (plot 2550) were used, using plasma methylome data of 4 healthy control patients as a reference to chromosome 1. Show. FIG. 2B is a table 2600 showing data for Z values for pre- and post-operative plasma. Circos plot 2620 showing Z values of plasma DNA methylation densities of pre-operative and post-operative plasma samples of HCC patients, using 4 healthy control patients as a reference for 1 Mb bins analyzed from all autosomes. . Table 2640 shows the distribution of whole genome 1 Mb bin Z values in both pre- and post-surgical plasma samples of HCC patients. Table 2660 shows methylation levels of tumor tissue and pre-operative plasma samples that overlap some of the control plasma samples with the CHH and CHG sequences. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. 6 shows a Circos plot of methylation densities of eight cancer patients, according to an embodiment of the invention. FIG. 3 is a table 2780 showing the number of sequence reads and the sequencing depth achieved for each sample. Table 2790 shows the Z value distribution of 1 Mb bins across the genome in plasma of patients with different malignancies. CL = lung adenocarcinoma; NPC = nasopharyngeal cancer; CRC = colorectal cancer; NE = neuroendocrine cancer; SMS = leiomyosarcoma. 6 is a flowchart of a method 2800 of analyzing a biological sample of an organism to determine a cancer level classification in accordance with an embodiment of the present invention. FIG. 6 is a plot 2900 showing the distribution of methylation densities in a reference subject, assuming the distribution follows a normal distribution. 3 is a plot 2950 showing the distribution of methylation densities in cancer subjects, assuming that the distribution is normal and the average methylation level is 2 standard deviations below the cutoff. 3 is a plot 3000 showing the distribution of plasma DNA methylation density in healthy subjects and cancer patients. FIG. 3B is a graph 3100 showing the distribution of differences in methylation density between plasma DNA of healthy subjects and mean values of tumor tissues of HCC patients. Table 3200 showing the effect of decreasing sequencing depth when plasma samples contained 5% or 2% tumor DNA. FIG. 3B is a graph 3250 showing the methylation densities of repetitive elements and non-repetitive regions of plasma from four healthy control subjects and buffy coat, normal liver tissue, tumor tissue, pre- and post-operative plasma samples of HCC patients. is there. FIG. 13 shows a block diagram of an exemplary computer system 3300 that can be used with the systems and methods according to embodiments of the invention. 3 shows the size distribution of plasma DNA in systemic lupus erythematosus (SLE) patient SLE04. 34 shows methylation analysis of plasma DNA obtained from SLE patient SLE04 (FIG. 34B) and HCC patient TBR36 (FIG. 34C). 34 shows methylation analysis of plasma DNA obtained from SLE patient SLE04 (FIG. 34B) and HCC patient TBR36 (FIG. 34C). 6 is a flowchart of a method 3500 for determining a cancer level classification based on CG island hypermethylation, in accordance with an embodiment of the present invention. 11 is a flow chart of a method 3600 of analyzing a biological sample of an organism using multiple chromosomal regions in accordance with an embodiment of the present invention. 3 shows CNA analysis (inside to outside) of tumor tissue, non-bisulfite (BS) -treated and bisulfite-treated plasma DNA of patient TBR36. FIG. 6 is a scatter plot showing the association between Z values in the detection of CNA using 1 Mb bin bisulfite-treated and non-bisulfite-treated plasma of patient TBR36. 3 shows CNA analysis of patient TBR34 tumor tissue, non-bisulfite (BS) -treated plasma DNA and bisulfite-treated plasma DNA (from inside to outside). FIG. 9 is a scatter plot showing the association between Z values in the detection of CNA using 1 Mb bin bisulfite-treated plasma and non-bisulfite-treated plasma of patient TBR34. FIG. 6 is a Circos plot showing CNA (inner ring) and methylation analysis (outer ring) of bisulfite treated plasma of HCC patient TBR240. FIG. 6 is a Circos plot showing CNA (inner ring) and methylation analysis (outer ring) of bisulfite treated plasma of HCC patient TBR164. 9 shows CNA analysis for patient TBR36 on pre-treatment and post-treatment samples. 14 shows methylation analysis for pre-treatment and post-treatment samples for patient TBR36. 9 shows CNA analysis for patient TBR34 on pre-treatment and post-treatment samples. 14 shows methylation analysis for pre-treatment and post-treatment samples for patient TBR34. Figure 4 shows a diagram of the diagnostic capabilities of genome-wide hypomethylation analysis using different numbers of sequenced reads. FIG. 6 shows ROC curves in cancer detection based on genome-wide hypomethylation analysis using different bin sizes (50 kb, 100 kb, 200 kb and 1 Mb). Shows the diagnostic ability of cumulative probability (CP) and the percentage of bins with abnormalities. Figure 3 shows the diagnostic ability of plasma analysis of global hypomethylation, CG island hypermethylation and CNA. FIG. 7 shows a table with global hypomethylation, CG island hypermethylation and CNA results in hepatocellular carcinoma patients. 1 shows a table containing global hypomethylation, CG island hypermethylation and CNA results in patients with cancers other than hepatocellular carcinoma. 12 shows a serial analysis of plasma methylation of case TBR34. FIG. 6 shows a Circos plot showing CNA (inner ring) and methylation changes (outer ring) in bisulfite treated plasma DNA of HCC patient TBR36. FIG. 6 is a plot of methylated Z-values of HCC patient TBR36 in regions with increased and decreased chromosomes, and in regions with unchanged copy number. FIG. 7 shows a Circos plot showing CNA (inner ring) and methylation changes (outer ring) in bisulfite treated plasma DNA of HCC patient TBR34. FIG. 9 is a plot of methylated Z-values of HCC patient TBR34 in regions with chromosomal gains and losses and in unchanged copy number. FIG. 6 shows the results of plasma hypomethylation and CNA analysis for SLE patients SLE04 and SLE10. FIG. 6 shows the results of plasma hypomethylation and CNA analysis for SLE patients SLE04 and SLE10. Z _meth analysis for plasma with and without CNA in plasma of two HCC patients (TBR34 and TBR36). Z _meth analysis for plasma with and without CNA in plasma of two HCC patients (TBR34 and TBR36). Z _meth analysis for the regions with and without CNA in the plasma of two SLE patients (SLE04 and SLE10). Z _meth analysis for the regions with and without CNA in the plasma of two SLE patients (SLE04 and SLE10). FIG. 7 shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients using group A features for CNA, global methylation, and CpG island methylation. 13 shows hierarchical clustering using group B features for CNA, global methylation, and CpG island methylation. FIG. 6 shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients using Group A CG island methylation features. FIG. 7 shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using Group A global methylation densities. FIG. 6 shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using Group A whole CNAs. FIG. 6 shows hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using Group B CpG island methylation density. 3 shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients. FIG. 6 shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using Group B global methylation densities. Shown is the average methylation density of 1 Mb bins (red dots) among 32 healthy subjects.

Definitions "Methylome" gives a measure of the amount of DNA methylation at multiple sites or loci within the genome. The methylome can correspond to the entire genome, a substantial portion of the genome, or a relatively small portion of the genome. “Fetal methylome” corresponds to the methylome of the fetus of a pregnant woman. Fetal methylome can be determined using various sources of fetal tissue or fetal DNA, such as placental tissue and cell-free fetal DNA in maternal plasma. "Tumor methylome" corresponds to a methylome of a tumor of an organism (eg, human). Tumor methylome can be determined using cell-free tumor DNA in tumor tissue or maternal plasma. Fetal and tumor methylomes are examples of methylomes of interest. Other examples of methylomes of interest are body fluids (eg, plasma, serum, sweat, saliva, urine, genital secretions, semen, stools fluid, diarrheal fluid, cerebrospinal fluid, gastrointestinal secretions. Substances, pancreatic secretions, intestinal secretions, sputum, tears, aspirates from the breast, and thyroid gland), and the methylomes of organs (eg, brain cells, bones, lungs, hearts, muscles and kidneys). Is. The organ may be a transplanted organ.

  A "plasma methylome" is a methylome determined from the plasma or serum of an animal (eg, human). Plasma methylome is an example of a cell-free methylome because plasma and serum contain cell-free DNA. Plasma methylome is also an example of a mixed methylome as it is a mixture of fetal / maternal methylome or tumor / patient methylome. “Placenta methylome” can be determined from a chorionic villus sample (CVS) or a placental tissue sample (eg, obtained after delivery). "Cellular methylome" corresponds to a methylome determined from cells (eg, blood cells) of a patient. Blood cell methylomes are referred to as blood cell methylomes (or blood methylomes).

  A “site” corresponds to a single site, which may be a single base position or a group of related base positions (eg, CpG sites). The “sitting position” corresponds to a region including a plurality of parts. A locus can include only one site that equates the locus to a site within its sequence.

  The "methylation index" for each genomic site (eg, CpG site) refers to the ratio of sequence reads that show methylation at the site to the total number of reads that cover that site. The "methylation density" of a region is the number of reads at a site within the region that exhibits methylation divided by the total number of reads covering the site within the region. The site may have a particular characteristic and may be, for example, a CpG site. Therefore, the "CpG methylation density" of a region is the CpG methylation divided by the total number of reads covering CpG sites within that region (eg, a particular CpG site, a CpG site within a CG island, or a larger region). Is the number of leads. For example, the methylation density of each 100 kb bin in the human genome is the cytosine unconverted after bisulfite treatment at the CpG site as the percentage of total CpG sites covered by the sequence reads located in the 100 kb region. Can be determined from the total number of (corresponding to methylated cytosine). This analysis can also be performed for other bin sites (eg 50 kb or 1 Mb, etc.). A region can be an entire genome or a chromosome or a portion of a chromosome (eg, chromosomal arm). The methylation index of a CpG site is the same as the methylation density of the region when the region only contains that CpG site. "Ratio of methylated cytosines" is shown to be methylated relative to the total number of cytosine residues analyzed (ie, including cytosines outside the CpG sequence within the region) (eg, bisulfite). Refers to the number of cytosine sites (“C”), not converted after salt conversion. The methylation index, methylation density and ratio of methylated cytosines are examples of "methylation levels".

  "Methylation properties" (also referred to as methylation status) contain information related to the DNA methylation of a region. The information relating to DNA methylation includes, but is not limited to, the methylation index of CpG sites, the methylation density of CpG sites within a region, the distribution of CpG sites over adjacent regions, and the region containing two or more CpG sites. Patterns or levels of methylation of individual CpG sites within and non-CpG methylation may be included. The methylation properties of a significant portion of the genome can be considered equivalent to methylome. “DNA methylation” in the mammalian genome typically refers to the addition of a methyl group to the 5 ′ carbon of cytosine residues between CpG dinucleotides (ie, 5-methylcytosine). DNA methylation can occur in cytosines within other sequences, such as CHG and CHH, where H is adenine, cytosine or thymine. Cytosine methylation may be in the form of 5-hydroxymethylcytosine. Non-cytosine methylation such as N6-methyladenine has also been reported.

  "Tissue" corresponds to all cells. Different types of tissue may correspond to different types of cells (eg liver, lung, or blood), but also to tissues obtained from different organisms (mother vs. fetus) or normal cells vs. tumor cells. A “biological sample” is a subject (eg, a human, eg, a pregnant woman, a cancer-bearing patient, or a suspected cancer-bearer, an organ transplant recipient, or an organ (eg, a heart in myocardial infarction, or a brain in stroke). Subject) suspected of having a disease hypothesis for) and containing one or more nucleic acid molecules of interest. Biological samples include body fluids such as blood, plasma, serum, urine, vaginal fluid, uterus or flushing fluid, multiple body fluids, ascites, cerebrospinal fluid, saliva, sweat, tears, sputum, bronchoalveolar lavage fluid, etc. It can be a body fluid. A stool sample can also be used.

  The term "cancer level" refers to the presence or absence of cancer, the stage of the cancer, the size of the tumor, the presence or absence of metastases, the total tumor burden of the body, and / or other measures of cancer severity. obtain. The level of cancer can be number or other characteristic. The level may be zero. The level of cancer also includes a premalignant or precancerous condition associated with a mutation or some mutations. Cancer levels can be used in a variety of ways. For example, screening can determine if cancer is present in a person previously unknown to have the cancer. By assessing who has been diagnosed with cancer by evaluation, it is possible to monitor the progression of cancer over time, study the efficacy of treatment, or determine prognosis. In one embodiment, the prognosis may represent the probability that the patient will die of cancer, or the cancer will progress after a certain period or time, or the cancer will metastasize. Detection can mean "screening" or can mean checking whether a person with characteristics suggestive of cancer (eg, symptoms or other positive tests) has cancer.

DETAILED DESCRIPTION Epigenetic machinery plays an important role in embryonic and fetal development. However, human embryonic and fetal tissues, including placental tissues, are not readily available (US Pat. No. 6,927,028). Certain embodiments addressed this issue by analyzing samples containing cell-free fetal DNA molecules present in the maternal circulation. The fetal methylome can be estimated in various ways. For example, maternal plasma methylome can be compared to cellular methylome (from maternal blood cells) and the difference has been shown to correlate with fetal methylome. As another example, fetal-specific alleles can be used to determine methylation of fetal methylome at a particular locus. Furthermore, the size of the fragment can be used as an indicator of the methylation rate, since a correlation between size and methylation rate is shown.

  In one embodiment, genome-wide bisulfite sequencing is used to analyze the methylation properties of maternal plasma DNA (some or all of the methylome) in single nucleotide resolution. By taking advantage of polymorphic differences between mother and fetus, the fetal methylome can be constructed from a maternal blood sample. In another implementation, polymorphic differences were used, but differences between plasma and blood cell methylomes can be used.

  In another embodiment, having cancer by utilizing single nucleotide mutations and / or copy number abnormalities between tumor and non-tumor genomes and sequencing data obtained from plasma (or other samples) Methylation profiling of tumors can be performed on samples of patients suspected of having cancer or diagnosed with cancer. The difference in the methylation level in the plasma sample of the test individual as compared to the plasma methylation level of the healthy control or the healthy control group may allow identification of the test individual as having cancer. In addition, methylation properties can serve as a signature that reveals the type of cancer, such as from which organ the patient developed and which metastasis originated.

  The non-invasive nature of this approach allowed the continuous evaluation of fetal and maternal plasma methylomes from maternal blood samples taken during the first, third and post-partum period. Changes related to pregnancy were observed. The approach can also be adapted to samples obtained during the second phase. The fetal methylome deduced from maternal plasma during pregnancy was similar to the placental methylome. Imprinted genes and differentially methylated regions were identified from maternal plasma data.

  By non-invasively, serially and comprehensively studying the fetal methylome, an approach has been developed that offers the possibility to identify biomarkers or directly test for pregnancy-related pathologies. The examples are also used to non-invasively study tumor methylome to screen or detect whether a subject has a kan, monitor malignancy in cancer patients, and for prognosis. , Continuous and comprehensive research is possible. Examples include, but are not limited to, lung cancer, breast cancer, colorectal cancer, prostate cancer, nasopharyngeal cancer, gastric cancer, testicular cancer, skin cancer (eg, melanoma), nerves. Systemic cancer, bone cancer, ovarian cancer, liver cancer (eg hepatocellular carcinoma), hematopoietic cancer, pancreatic cancer, endometriocarcinoma, renal cancer, cervical cancer, It can be applied to bladder cancer and the like.

  A description of methods for determining methylome or methylation properties is first considered, and then various methylomes (eg, fetal methylome, tumor methylome, mother or patient methylome, and mixed methylomes (eg, obtained from plasma) ) Is described. Next, the determination of fetal methylation properties using fetal-specific markers or by comparing mixed methylation properties with cellular methylation properties is described. Fetal methylation markers are determined by comparing methylation profiles. Consider the relationship between size and methylation. Also provided is the use of methylation properties to detect cancer.

I. Determining the Methylome A myriad of approaches have been used to study placental methylome, but each approach has its limitations. For example, sodium bisulfite is a chemical that changes unmethylated cytosine residues to uracil and does not change methylated cytosine, which translates differences in cytosine methylation into gene sequence differences for further matching. And convert. The gold standard method of studying cytosine methylation is based on treating tissue DNA with sodium bisulfite followed by direct sequencing of individual clones of the bisulfite converted DNA molecule. After analyzing multiple clones of a DNA molecule, the pattern and quantitative characteristics of cytosine methylation per CpG site can be obtained. However, cloned bisulfite sequencing is a low-throughput and very laborious procedure that cannot be easily applied on a genome-wide scale.

  Methylation sensitive restriction enzymes, which typically digest unmethylated DNA, provide an inexpensive approach to study DNA methylation. However, the data obtained from such studies are limited to loci with enzyme recognition motifs and the results are non-quantitative. Immunoprecipitation of DNA bound by anti-methylated cytosine antibodies can be used to probe large segments of the genome, but such antibodies are more strongly bound to loci with high density of methylation. Tends to be biased in the area. Microarray-based approaches rely on the a priori design of reference probes and the efficiency of hybridization between the probes and target DNA.

  To comprehensively examine the methylome, in some embodiments, massively parallel sequencing (MPS) is used to provide genome-wide information on methylation levels per nucleotide and per allele. And a quantitative assessment is given. Recently, genome-wide MPS after bisulfite conversion has become feasible (R Lister et al 2008 Cell; 133: 523-536).

  A small number of published studies applying genome-wide bisulfite sequencing to investigate human methylomes (R Lister et al. 2009 Nature; 462: 315-322; L Laurent et al. 2010 Genome Res; 20: 320-331. Y Li et al. 2010 PLoS Biol; 8: e1000533; and M Kulis et al. 2012 Nat Genet; 44: 1236-1242), two studies focused on embryonic stem cells and fetal fibroblasts. (R Lister et al. 2009 Nature; 462: 315-322; L Laurent et al. 2010 Genome Res; 20: 320-331). Both of these studies analyzed DNA from cell lines.

A. Genome-Wide Bisulfite Sequencing Certain embodiments are capable of overcoming the above problems and allow fetal methylome matching to be comprehensive, non-invasive, and continuous. In one embodiment, genome-wide bisulfite sequencing was used to analyze cell-free fetal DNA molecules present in the bloodstream of pregnant women. Since the plasma DNA molecules were fragmented in small amounts, it was possible to continuously observe changes with the progress of pregnancy by constructing a high resolution fetal methylome from maternal plasma. In light of the strong interest in noninvasive prenatal testing (NIPT), embodiments provide powerful new means for discovering fetal biomarkers, or fetal-related diseases or pregnancy. It can serve as a direct platform for achieving NIPT of related diseases. Data obtained from genome-wide bisulfite sequencing of various samples from which fetal methylome can be obtained is provided below. In one embodiment, this technology can be applied to methylation profiling in preeclampsia, or in utero fetal growth retardation, or pregnancy associated with preterm birth. In such combined pregnancies, this technology can be used continuously due to its non-invasive nature to enable monitoring and / or prognosis and / or response to treatment.

  FIG. 1A shows the results of Table 100 of maternal blood, placenta, and maternal plasma sequencing according to embodiments of the present invention. In one embodiment, whole-genome sequencing was performed on blood cells of a blood sample collected at the first trimester, CVS, placental tissue collected at the end of the trimester, first and third trimesters, and puerperium. Against a bisulfite converted DNA library of a maternal plasma sample prepared using a methylated DNA library adapter (Illumina) (R Lister et al. 2008 Cell; 133: 523-536). ,It was conducted. Blood cell and plasma DNA samples from one adult male and one adult non-pregnant female were also analyzed. A total of 9.5 billion pairs of live sequence reads were generated in this study. The sequencing coverage of each sample is shown in Table 100.

  Sequence reads that were uniquely mappable to the human reference genome reached mean haploid genome coverages of 50-fold, 34-fold and 28-fold, respectively, in first-stage, third-stage and post-natal maternal plasma samples. . The coverage of CpG sites in the genome ranged from 81% to 92% in the samples obtained from the gestation period. Sequence reads spanning the CpG site reached an average haploid coverage of 33-fold / chain, 23-fold / chain and 19-fold / chain in the first, third, and postnatal maternal plasma samples, respectively. The bisulfite conversion efficiency of all samples was above 99.9% (Table 100).

  In Table 100, the ambiguous percentage (labeled "a") refers to the percentage of reads that were mapped to both the Watson and Crick strands of the reference human genome. Lambda conversion refers to the percentage of unmethylated cytosines in the internal Lambda DNA control that have been converted to "thymine" residues by bisulfite modification. H is generally equal to A, C, or T. "A" refers to a lead that can be mapped to a particular genomic locus but cannot be assigned to the Watson or Crick strands. "B" refers to lead pairs that have the same start and end coordinates. For "c", lambda DNA was spiked into each sample prior to bisulfite conversion. The lambda conversion rate refers to the proportion of cytosine nucleotides remaining as cytosines after bisulfite conversion and is used as an indicator of the proportion of successful bisulfite conversion. "D" refers to the number of cytosine nucleotides that are present in the reference human genome and remain as cytosine sequences after bisulfite conversion.

  During bisulfite modification, unmethylated cytosine is converted to uracil and then to thymine after PCR amplification, while methylated cytosine remains unchanged (M Frommer et al. 1992 Proc Natl Acad. Sci USA; 89: 1827-31). Therefore, after sequencing and alignment, the methylation status of individual CpG sites is determined by the number of methylated sequence reads "M" (methylated) and unmethylated sequence lead "U" at cytosine residues within the CpG sequence. It can be inferred from the number of (unmethylated). Using bisulfite sequencing data, a whole methylome of maternal blood, placenta and maternal plasma was constructed. The average methylated CpG density (also referred to as methylated density MD) of a specific locus in maternal plasma can be calculated using the following formula:

Where M is the number of methylated reads at the CpG site within the locus and U is the number of unmethylated reads at the CpG site within the locus. If there is more than one CpG site within the locus, and U corresponds to the total number of those sites.

B. Various approaches As described above, methylation profiling can be performed using massively parallel sequencing (MPS) of plasma DNA converted by bisulfite. MPS of plasma DNA converted by bisulfite can be performed in a random or shotgun fashion. Sequencing depth can vary depending on the size of the region of interest.

  In another embodiment, the region of interest within the plasma DNA converted by bisulfite can be first captured using a solution or solid phase hybridization based process followed by MPS. Large-scale parallel sequencing includes sequencing-by-synthesis platform by synthesis (eg Illumina), sequencing-by-ligation platform by ligation (eg SOLiD platform manufactured by Life Technologies), semiconductor Performing using a base-based sequencing system (eg, Ion Torrent or Ion Proton platform from Life Technologies) or a single molecule sequencing system (eg, Helicos system or Pacific Biosciences system or nanopore-based sequencing system). You can Nanopore-based sequencing includes, for example, nanopores constructed with lipid bilayers and protein nanopores, as well as solid-state nanopores (eg, graphene-based nanopores). Selected single-molecule sequencing platforms can directly elucidate the methylation status of DNA molecules (eg, N6-methyladenine, 5-methylcytosine and 5-hydroxymethylcytosine) without bisulfite conversion. (BA Flusberg et al. 2010 Nat Methods; 7: 461-465; J Shim et al. 2013 Sci Rep; 3: 1389. Doi: 10.1038 / srep01389). It allows analysis of the methylation status of sample DNA that has not been converted by salt (eg plasma DNA).

  In addition to sequencing, other techniques can be used. In one embodiment, methylation profiling can be performed by methylation specific PCR or methylation sensitive restriction enzyme digestion followed by PCR or ligase chain reaction followed by PCR. In yet another embodiment, PCR is in the form of single molecule or digital PCR (B Vogelstein et al. 1999 Proc Natl Acad Sci USA; 96: 9236-9241). In yet another embodiment, the PCR can be real-time PCR. In other embodiments, the PCR can be multiplex PCR.

II. Methylome analysis In some embodiments, whole genome bisulfite sequencing can be used to determine the methylation profile of plasma DNA. Fetal methylation properties can be determined by sequencing maternal plasma DNA samples as described below. Therefore, fetal DNA molecules (and fetal methylome) were non-invasively accessible during pregnancy and changes were continuously monitored as pregnancy progressed. Due to the comprehensiveness of the sequencing data, it was possible to study the maternal plasma methylome on a genome-wide scale at single nucleotide resolution.

  Since the genomic location information of the sequenced reads was known, these data allowed the study of global methylation levels of the methylome or any region of interest within the genome, allowing comparisons between different genetic elements. . In addition, multiple sequence reads covered each CpG site or locus. Below is a description of some of the metrics used to measure methylome.

A. Methylation of Plasma DNA Molecules DNA molecules are present in human plasma in low concentrations, in fragmented form, typically similar in length to mononucleosome units (YMD Lo et al. 2010 Sci. Transl Med; 2: 61ra91; and YW Zheng at al. 2012 Clin Chem; 58: 549-558). Despite these limitations, the genome-wide bisulfite sequencing pipeline was able to analyze the methylation of plasma DNA molecules. In yet another embodiment, the single molecule sequencing platform of choice allows direct elucidation of the methylation status of DNA molecules without bisulfite conversion (BA Flusberg et al. 2010 Nat Methods 7: 461-465; J Shim et al. 2013 Sci Rep; 3: 1389. Doi: 10.1038 / srep01389), the use of such a platform allows the conversion of plasma DNA not converted by bisulfite into plasma DNA. It can be used for measuring methylation levels or for measuring plasma methylome. Such platforms may provide improved results (eg, improved sensitivity or specificity) associated with different biological functions of different forms of methylation, N6-methyladenine, 5-methylcytosine, and 5 -Hydroxymethylcytosine can be detected. Such improved outcomes may be useful when applying the embodiments to the detection or monitoring of particular disorders (eg, preeclampsia) or particular cancer types.

  Bisulfite sequencing can also identify different forms of methylation. In one embodiment, bisulfite sequencing may include an additional step capable of distinguishing 5-methylcytosine from 5-hydroxymethylcytosine. One such approach is oxidative bisulfite sequencing (oxBS-seq), which locates 5-methylcytosine and 5-hydroxymethylcytosine at single base resolution. It can be clarified (MJ Booth et al. 2012 Science; 336: 934-937; MJ Booth et al. 2013 Nature Protocols; 8: 1841-1851). Bisulfite sequencing is indistinguishable because both 5-methylcytosine and 5-hydroxymethylcytosine are read as cytosines. On the other hand, in oxBS-seq, treatment of 5-hydroxymethylcytosine by treatment with potassium perruthenate (KRuO4) followed by conversion of newly formed 5-formylcytosine to uracil using bisulfite conversion. The specific oxidation to formylcytosine makes it possible to distinguish 5-hydroxymethylcytosine from 5-methylcytosine. Therefore, a 5-methylcytosine reading can be obtained from a single oxBS-seq run, and 5-hydroxymethylcytosine levels are estimated by comparing the bisulfite sequencing results. In another embodiment, 5-methylcytosine can be distinguished from 5-hydroxymethylcytosine by using Tet-assisted bisulfite sequencing (TAB-seq) (M. Yu et al. 2012 Nat Protoc; 7: 2159-2170). TAB-seq can specify 5-hydroxymethylcytosine with single base resolution, and can determine its abundance at each modification site. This method involved β-glucosyltransferase-mediated protection of 5-hydroxymethylcytosine (glucosylation), and recombinant mouse Tet1 (mTet1) -mediated oxidation of 5-methylcytosine to 5-carboxylcytosine. included. After subsequent bisulfite treatment and PCR amplification, both cytosine and 5-carboxyl cytosine (from 5-methylcytosine) are converted to thymine (T), while 5-hydroxymethylcytosine is read as C.

  FIG. 1B shows the methylation density within the 1 Mb window of samples sequenced according to embodiments of the invention. Plot 150 is a Circos plot showing the methylation density in maternal plasma and genomic DNA within the 1 Mb window across the genome. From outside to inside: Chromosome schematic can be in the pter-qter orientation in the clockwise direction: (centromere shown in red), maternal blood (red), placenta (yellow), maternal plasma (green). , Shared lead in maternal plasma (blue), and fetal-specific lead in maternal plasma (purple). Total CpG methylation levels (ie, density levels) in maternal blood cells, placenta and maternal plasma can be found in Table 100. Maternal blood cell methylation levels are usually higher than placental methylation levels across the entire genome.

B. Comparison of bisulfite sequencing and other techniques Placental methylome was studied using massively parallel bisulfite sequencing. In addition, placental methylomes were studied using an oligonucleotide array platform (Illumina) that covers approximately 480,000 CpG sites in the human genome (M Kulis et al. 2012 Nat Genet; 44: 1236-1242; and C Clark). et al. 2012 PLoS One; 7: e50233). In one embodiment using bead-chip based genotyping and methylation analysis, genotyping was performed using the Illumina Human Omni2.5-8 genotyping array according to the manufacturer's protocol. Genotypes were called using the GenCall algorithm of Genome Studio Software (Illumina). The call rate was over 99%. In a microarray-based methylation analysis, genomic DNA (500-800 ng) was loaded into a Zymo EZ DNA methylation kit (Zymo Research, Calif., USA) for Illumina Infinity Methylation assay according to manufacturer's recommendations. Orange) and treated with sodium bisulfite.

  Methylation assays were performed on 4 μl of bisulfite converted genomic DNA (50 ng / μl) according to the Infinium HD methylation assay protocol. The hybridized bead chips were scanned on an IlScan iScan device. DNA methylation data was analyzed by GenomeStudio (v2011.1) Methylation Module (v1.9.0) software, normalization to an internal standard and background removal were also performed. The methylation index of an individual CpG site is represented by the β value (β), which is calculated using the ratio of fluorescence intensity between the methylated and unmethylated alleles:

  For CpG sites represented on the array and sequenced for at least 10-fold coverage, the β value obtained by the array was compared to the methylation index determined by sequencing the same sites. The β value represents the intensity of the methylated probe as a ratio of the mixed intensities of the methylated probe and the unmethylated probe covering the same CpG site. The methylation index of each CpG site refers to the ratio of methylated reads to the total number of reads covering that CpG.

2A-2C are β-values determined by Illumina Infinium HumanMethylation 450K bead chip array versus methylation index determined by genome-wide bisulfite sequencing of the corresponding CpG sites probed by both platforms. Shows plots of: (A) maternal blood cells, (B) chorionic villous specimen, (C) late pregnancy placental tissue. The data obtained from both platforms are highly consistent, with Pearson correlation coefficients of 0.972, 0.939 and 0.954 for maternal blood cells, CVS and late pregnancy placental tissue, respectively, and R ² The values were 0.945, 0.882 and 0.910.

In addition, the sequencing data was compared to the sequencing data reported by Chu et al, which examined the methylation properties of 12 pairs of CVS and maternal blood cell DNA samples using an oligonucleotide array covering approximately 27,000 CpG sites. (T Chu et al. 2011 PLoS One; 6: e14723). Correlation data between CVS and maternal blood cell DNA and the sequencing results of each of the 12 pairs of samples in the previous study yielded mean Pearson coefficient (0.967) and R ² (0.935) in maternal blood. , CVS average Pearson counts (0.943) and R ² (0.888) were obtained. Of the CpG sites represented on both arrays, our data were highly correlated with published data. The percentage of non-CpG methylation in maternal blood cells, CVS and placental tissues was less than 1% (Table 100). These results were consistent with the current finding that significant amounts of non-CpG methylation were mainly restricted to pluripotent cells (R Lister et al. 2009 Nature; 462: 315-322; L Laurent et al. 2010 Genome Res; 20: 320-331).

C. Comparison of Plasma Methylome and Blood Methylome in Non-Pregnant Subjects FIGS. 3A and 3B show a bar graph of the percentage of methylated CpG sites in plasma and blood cells collected from adult males and non-pregnant adult females: (A) Autosome, (B) X chromosome. The chart shows the similarities between plasma and blood methylomes in males and non-pregnant women. The overall proportion of methylated CpG sites in male and non-pregnant female plasma samples was about the same as the corresponding blood cell DNA (Table 100 and Figures 2A and 2B).

  Next, the locus-specific correlation of the methylation properties of plasma and blood cell samples was studied. The methylation density of each 100 kb bin in the human genome was determined by determining the total number of unconverted cytosines at the CpG sites as a percentage of the total CpG sites covered by the sequence reads mapped to the 100 kb region. The methylation density was highly consistent between plasma samples of male and female samples and the corresponding blood cell DNA.

Figures 4A and 4B show plots of methylation density of corresponding loci in blood cell and plasma DNA: (A) non-pregnant adult female, (B) adult male. Pearson correlation coefficient and R ² value of the non-pregnant female samples are each 0.963 and 0.927, Pearson correlation coefficient and R ² value of male samples were, respectively, 0.953 and 0.908. These data indicated that hematopoietic cells were the major source of DNA in human plasma, with previous findings based on the genotyping of plasma DNA molecules in recipients of allogeneic hematopoietic stem cell transplantation (YW Zheng at al. 2012 Clin Chem; 58: 549-558).

D. Methylome-wide methylation levels Next, DNA methylation levels of maternal plasma DNA, maternal blood cells, and placental tissues were examined to determine methylation levels. Levels in repetitive, non-repetitive and global regions were determined.

  5A and 5B show bar graphs of the percentage of methylated CpG sites among samples taken from pregnant women: (A) autosomal chromosome, (B) X chromosome. The total percentage of methylated CpGs was 67.0% and 68.2% in the first and third stage maternal plasma samples, respectively. Unlike the results obtained from non-pregnant individuals, these rates were lower than that of the first trimester maternal blood cell sample, but higher than that of the CVS and late pregnancy placental tissue specimens (Table 100). Of note, the percentage of methylated CpG in the postnatal maternal plasma sample was 73.1%, which was similar to the blood cell data (Table 100). These trends were observed in CpG, which is distributed across all autosomal and X chromosomes, spanning both the non-repetitive regions of the human genome and multiple classes of repetitive regions.

  Both repetitive and non-repetitive regions in the placenta were found to be hypomethylated compared to maternal blood cells. These results were consistent with findings in the literature that placenta is hypomethylated compared to other tissues (eg, peripheral blood cells).

  In blood cell DNA obtained from pregnant women, non-pregnant women and adult men, 71% to 72% of the sequenced CpG sites were methylated (Table 100 in Figure 1). These data are comparable to the report of 68.4% of CpG sites in blood mononuclear cells reported by Y Li et al. 2010 PLoS Biol; 8: e1000533. Consistent with previous reports on the hypomethylated nature of placental tissue, 55% and 59% of CpG sites were methylated in CVS and late pregnancy placental tissue, respectively (Table 100).

  FIG. 6 shows a bar graph of methylation levels of different repeat classes of the human genome in maternal blood, placenta and maternal plasma. Repeat classes are defined by the UCSC Genome Browser. The data shown are from the first trimester samples. Unlike the earlier data (B Novakovic et al. 2012 Placenta; 33: 959-970), which showed that the hypomethylated nature of placental tissues was mainly observed in a particular recurrent class within the genome, where placenta Show that they were actually hypomethylated in most classes of genomic elements, relative to blood cells.

E. Methylome Affinity Embodiments allow the same platform to be used to determine methylome in placental tissue, blood cells and plasma. Therefore, it was possible to directly compare the methylomes of those biological sample types. The high level of similarity between blood cell and plasma methylomes in males and non-pregnant women, as well as between maternal blood cells and postnatal maternal plasma samples further confirms that hematopoietic cells are the major source of DNA in human plasma. (YW Zheng at al. 2012 Clin Chem; 58: 549-558).

  The similarity is clear in terms of the total percentage of methylated CpG within the genome and from the high correlation of methylation density between corresponding loci in blood cell DNA and plasma DNA. However, the total proportion of methylated CpG in the first and third stage maternal plasma samples was reduced when compared to maternal blood cell data or postnatal maternal plasma samples. The reduction in methylation levels during pregnancy was due to the hypomethylated nature of fetal DNA molecules present in maternal plasma.

  The reversal of the methylation profile in postnatal maternal plasma samples to make them more similar to the methylation profile of maternal blood cells suggests that fetal DNA molecules have been eliminated from the maternal circulation. Calculation of fetal DNA concentration based on the fetal SNP marker showed that the concentration did change from 33.9% prenatal to only 4.5% in postnatal samples.

F. Other Applications Embodiments have been successful in constructing DNA methylomes through MPS analysis of plasma DNA. The ability to determine placental or fetal methylome from maternal plasma is a non-invasive property for determining, detecting and monitoring aberrant methylation properties associated with pregnancy-related conditions such as preeclampsia, intrauterine growth retardation and preterm birth. Provide a way. For example, detection of disease-specific aberrant methylation signatures allows for screening, diagnosis and monitoring of such pregnancy related conditions. Measurement of maternal plasma methylation levels allows for screening, diagnosis and monitoring of such pregnancy related conditions. In addition to its direct application to the study of pregnancy-related conditions, the prophase approach can be applied to other areas of medicine for plasma DNA analysis. For example, the cancer methylome can be determined from the plasma DNA of cancer patients. The cancer methylome analysis from plasma described herein can be a synergistic science and technology for cancer genome analysis from plasma (KCA Chan at al. 2013 Clin Chem; 59: 211-224. And RJ Leary et al. 2012 Sci Transl Med; 4: 162ra154).

  For example, the determination of methylation levels in plasma samples can be used to screen for cancer. Cancer may be suspected if the methylation level of the plasma sample shows abnormal levels compared to healthy controls. Subsequent determination of the cancer type or phenotype by determining plasma characteristics of methylation at different genomic loci or by plasma genomic analysis to detect tumor-associated copy number abnormalities, chromosomal translocations, and single nucleotide mutations. Further confirmation and evaluation of the tissue origin of cancer may be performed. Indeed, in one embodiment of the invention, the cancer methylome and genome profiling in plasma can be performed simultaneously. Alternatively, by using radiography and diagnostic imaging (eg, computed tomography, magnetic resonance imaging, positron emission tomography) or endoscopy (eg, upper gastrointestinal endoscopy or colonoscopy) Individuals suspected of having cancer can be further tested based on plasma methylation level analysis.

  In the screening or detection of cancer, the determination of methylation levels of plasma samples (or other biological samples) can be used in conjunction with other modalities for screening or detection of cancer, eg prostate specific antigen measurement. (For example, prostate cancer), carcinoembryonic antigen (for example, colorectal cancer, gastric cancer, pancreatic cancer, lung cancer, breast cancer, medullary thyroid cancer), α-fetoprotein (for example, liver cancer or germ cell tumor), CA125 (eg, for ovarian and breast cancer) and CA19-9 (eg, for pancreatic cancer).

  In addition, other tissues can be sequenced to obtain cellular methylomes. For example, analysis of liver tissue can determine liver-specific methylation patterns, which may be used to identify liver lesions. Other tissues that can be analyzed include brain cells, bones, lungs, heart, muscles and kidneys. Methylation properties of various tissues can be characterized, for example, as development, aging, disease hypothesis (eg, inflammation or cirrhosis or autoimmune processes (eg, systemic lupus erythematosus)), or treatment (eg, 5-azacytidine and 5-azacytidine. As a result of treatment with a demethylating agent such as deoxycytidine). The dynamic nature of DNA methylation makes such an analysis potentially very valuable in monitoring physiological and pathological processes. For example, a disease hypothesis in an organ that provides plasma DNA can be detected when detecting changes in an individual's plasma methylome compared to baseline values obtained when the individual was healthy.

  Further, the methylome of the transplanted organ can be determined from the plasma DNA of the organ transplant recipient. The graft methylome analysis from plasma described in the present invention can be a synergistic science and technology for the analysis of graft genome from plasma (YW Zheng at al, 2012; YMD Lo at al. 1998 Lancet; 351: 1329). -1330; and TM Snyder et al. 2011 Proc Natl Acad Sci USA; 108: 6229-6234). Since plasma DNA is generally regarded as a marker of cell death, an increase in plasma levels of DNA released from a transplanted organ results in an increase in cell death from that organ, eg, a rejection episode or the organ. It can be used as a marker for other pathological processes involved in (eg, infection or abscess). In the event that anti-rejection therapy is successfully initiated, plasma levels of DNA released by transplanted organs are expected to decrease.

III. Determination of Fetal or Tumor Methylome Using SNPs As described above, plasma methylome is consistent with blood methylome in non-pregnant healthy individuals. However, in pregnant women these methylomes are different. Fetal DNA molecules circulate in maternal plasma in most maternal DNA backgrounds (YMD Lo et al. 1998 Am J Hum Genet; 62: 768-775). Therefore, in pregnant women, the plasma methylome is predominantly a mixture of placental and blood methylomes. Therefore, the placenta methylome can be extracted from plasma.

  In one embodiment, single nucleotide polymorphism (SNP) differences between mother and fetus are used to identify fetal DNA molecules in maternal plasma. Aims to identify SNP loci in which the mother is homozygous but the fetus is heterozygous; using fetal-specific alleles to determine which DNA fragment is from the fetus You can Maternal blood cell-derived genomic DNA was analyzed using a SNP genotyping array, Human Omni 2.5-8 manufactured by Illumina. On the other hand, for SNP loci in which the mother is heterozygous and the fetus is homozygous, it is possible to determine which plasma DNA fragment is derived from the mother by using the SNP allele specific to the mother. it can. The methylation level of such DNA fragments reflects the methylation level of the relevant genomic region in the mother.

A. Correlation of fetal-specific lead methylation and placental methylome A locus with two different alleles is significantly less abundant in the amount of one allele (B) than in the other (A). It was identified from the results of sequencing the samples. Reads covering the B allele were considered fetal-specific (fetal-specific reads). Since it was determined that the mother was homozygous for A and the fetus was heterozygous for A / B, the lead covering the A allele was shared by the mother and the fetus (shared lead).

  In one analyzed case of pregnancy, used to explain some of the aspects of the invention, the pregnant mother is homozygous at the 1,945,516 locus on the autosomal chromosome. I understood. The maternal plasma DNA sequencing reads covering these SNPs were examined. Leads carrying non-maternal alleles were detected at 107,750 loci, which were considered to be informative loci. In each informative SNP, the non-maternal allele was designated as the fetal-specific allele and the other alleles as the shared allele.

  The fetal / tumor DNA concentration fraction in maternal plasma (also called fetal DNA percentage) can be determined. In one embodiment, the fetal DNA concentration fraction (f) in maternal plasma is determined by the formula:

Where p is the number of sequenced reads with the fetus-specific allele and q is the number of sequenced reads with the allele shared between the mother and the fetus (YMD Lo et al. 2010 Sci Transl Med; 2: 61ra91). The fetal DNA rates in the first, third and post-natal maternal plasma samples were found to be 14.4%, 33.9% and 4.5%, respectively. The fetal DNA ratio was calculated using the number of reads arranged on the Y chromosome. Based on Y chromosome data, the results were 14.2%, 34.9% and 3.7% in the 1 st, 3 rd and post-natal maternal plasma samples, respectively.

  By analyzing fetal-specific or shared sequence reads separately, embodiments show that circulating fetal DNA molecules were even more hypomethylated than background DNA molecules. Comparison of methylation densities of corresponding loci in fetal-specific maternal plasma leads and placental tissue data for both first and third trimesters revealed high levels of correlation. These data provide genomic-level evidence that the placenta is the major source of fetal-derived DNA molecules in maternal plasma, compared to earlier evidence based on information obtained from selected loci, It shows great progress.

  The methylation density of each 1 Mb region in the genome was determined using a fetal-specific or shared read covering the CpG site adjacent to the informative SNP. Fetal-specific and non-fetal-specific methylomes constructed from maternal plasma sequence reads can be displayed, eg, in the Circos plot (M Krzywinski et al. 2009 Genome Res; 19: 1639-1645). The methylation density per bin of 1 Mb in maternal blood cells and placental tissue specimens was also determined.

  FIG. 7A shows a Circos plot 700 of the first trimester sample. FIG. 7B shows a Circos plot 750 of the third trimester sample. Plot 700 and plot 750 show methylation density per bin of 1 Mb. The schematic diagram of the chromosome (outermost ring) is in the pter-qter orientation in the clockwise direction (centromere is shown in red). The second track from the outermost side shows the number of CpG sites in the corresponding 1 Mb region. The red bar scale shown is up to 20,000 sites per 1 Mb bin. The methylation density of the corresponding 1 Mb region is shown in the other tracks based on the color scheme shown in the center.

  In the first trimester sample (FIG. 7A), from the inside to the outside, the tracks are chorionic villous specimens, fetal-specific leads in maternal plasma, maternal-specific leads in maternal plasma, fetal leads in maternal plasma and A combination of non-fetal leads, as well as maternal blood cells. In the third trimester sample (FIG. 7B), tracks indicate late pregnancy placental tissue, fetal-specific reads in maternal plasma, maternal-specific reads in maternal plasma, fetal and non-fetal leads in maternal plasma, and delivery. Post maternal plasma as well as maternal blood cells (obtained from first stage blood samples). It can be seen that in both the first and third stage plasma samples, the fetal methylome was more hypomethylated than that of the non-fetal specific methylome.

  The overall methylation profile of fetal methylome was very similar to that of CVS or placental tissue specimens. On the contrary, the DNA methylation properties of the shared reads in plasma, which were predominantly maternal DNA, were very similar to those of maternal blood cells. Next, a systematic locus-by-locus comparison of maternal plasma DNA leads and maternal or fetal tissue methylation densities was performed. The methylation density of CpG sites that were present as informative SNPs on the same sequence read and covered by at least 5 maternal plasma DNA sequence reads was determined.

  8A-8D show a plot of the comparison of methylation densities of genomic tissue DNA to maternal plasma DNA for CpG sites surrounding the informative single nucleotide polymorphism. FIG. 8A shows the methylation density of fetal-specific reads in the first trimester maternal plasma sample compared to the methylation density of reads in the CVS sample. As shown in the figure, the value of the fetus-specific read is in good agreement with the value of the CVS read.

  FIG. 8B shows the methylation densities of fetal-specific reads in the third trimester maternal plasma sample compared to the methylation densities of reads for late pregnancy placental tissue. Once again, the densities of these sets are in good agreement, indicating that fetal methylation profiles can be obtained by analyzing reads with fetal-specific alleles.

  FIG. 8C shows the methylation densities of the shared reads in the first trimester maternal plasma sample compared to the methylation densities of the leads in maternal blood cells. These two sets of values are in good agreement, considering that the majority of the shared reads are of maternal origin. FIG. 8D shows the methylation density of shared leads in the third trimester maternal plasma sample compared to the methylation density of leads in maternal blood cells.

  In fetal-specific reads in maternal plasma, the Spearman correlation coefficient between first trimester maternal plasma and CVS is 0.705 (P <2.2 * e-16); third trimester maternal plasma and late pregnancy placenta The Spearman correlation coefficient between tissues was 0.796 (P <2.2 * e-16) (FIGS. 8A and 8B). Similar comparisons were made for maternal plasma shared reads and maternal blood cell data. The Pearson correlation coefficient is 0.653 (P <2.2 * e-16) in the first stage plasma sample, and the Pearson correlation coefficient is 0.638 (P <2.2 * e in the third stage plasma sample. -16) (FIGS. 8C and 8D).

B. Fetal Methylome In one embodiment, at least one informative fetal SNP site was coated and sequence reads containing at least one CpG site within the same read were selected to construct a fetal methylome from maternal plasma. Reads that showed a fetal-specific allele were included in the fetal methylome construction. Reads that showed shared alleles (ie, non-fetal-specific alleles) were included in the construction of non-fetal-specific methylomes that consist primarily of maternally-derived DNA molecules.

  The fetal-specific reads of the first trimester maternal plasma sample covered the 218,010 CpG sites on the autosomal chromosome. The corresponding numbers for the third trimester and post-natal maternal plasma samples were 263,611 and 74,020, respectively. On average, the shared reads averaged 33.3-fold, 21.7-fold, and 26.3-fold their CpG sites, respectively. Fetal-specific reads of the first, third and post-natal maternal plasma samples covered their CpG sites by 3.0-fold, 4.4-fold and 1.8-fold, respectively.

  Since fetal DNA is a minor population in maternal plasma, the coverage of their CpG sites by fetal-specific reads was proportional to the fetal DNA percentage of the sample. In the first trimester maternal plasma sample, the overall percentage of methylated CpG between fetal leads was 47.0%, while the overall percentage of shared lead methylated CpG was 68.1%. In the third trimester maternal plasma sample, the fetal lead had a methylated CpG ratio of 53.3%, while the shared lead had a methylated CpG ratio of 68.8%. These data showed that the fetal-specific reads in maternal plasma had a lower methylation status than the shared reads in maternal plasma.

C. Method Tumor methylation properties can also be determined using the above procedure. Described below are methods for determining fetal and tumor methylation properties.

  FIG. 9 is a flow chart illustrating a method 900 for determining a first methylation property from a biological sample of an organism, according to an embodiment of the invention. Method 900 can build an epigenetic map of the fetus from the methylation profile of maternal plasma. The biological sample includes cell-free DNA containing a mixture of cell-free DNA from the first tissue and the second tissue. By way of example, the first tissue can be obtained from a fetus, a tumor, or a transplanted organ.

  At block 910, a plurality of DNA molecules are analyzed from the biological sample. Analysis of DNA molecules includes determining the position of the DNA molecule within the genome of the organism, determining the genotype of the DNA molecule, and determining whether the DNA molecule is methylated at one or more sites. Can be included.

  In one embodiment, the DNA molecule is analyzed using a sequence read of the DNA molecule, where sequencing determines methylation. Therefore, the sequence reads include DNA molecule methylation status from the biological sample. The methylation status can include whether a particular cytosine residue is 5-methylcytosine or 5-hydroxymethylcytosine. Sequence reads can be obtained from a variety of sequencing methods, PCR methods, arrays, and other suitable techniques for identifying the sequence of fragments. The methylation status at the site of the sequence read can be obtained as described herein.

  At block 920, the first genome of the first tissue is heterozygous for each first allele and each second allele, and the second genome of the second tissue is homozygous for each first allele. , Multiple first loci are identified. For example, fetal-specific reads can be identified at multiple first loci. Alternatively, tumor-specific leads can be identified at multiple first loci. Tissue-specific reads have a percentage of sequence reads of the second allele falling within a particular range, eg, about 3% to 25%, which results in a small number of DNA fragments from the heterozygous genome at that locus. It can be identified from sequencing reads, which are populations and are shown to be the majority population of DNA fragments from the homozygous genome at that locus.

  At block 930, DNA molecules located at one or more sites in each first locus are analyzed. Several DNA molecules that are methylated at the site and correspond to each second allele at that locus are determined. There can be more than one site per locus. For example, a SNP may indicate that the fragment is fetal-specific and that the fragment may have multiple sites whose methylation status has been determined. The number of reads at each methylated site can be determined and the total number of methylated reads at that locus can be determined.

  A locus can be defined by a specific number of sites, a specific set of sites, or the specific size of the region surrounding a mutation containing a tissue-specific allele. The locus can have only one site. The site may have specific properties, for example a CpG site. The determination of the number of unmethylated reads is the same and is included in the determination of methylation status.

  At block 940, for each of the first loci, the methylation density is methylated at one or more sites of the locus, based on the number of DNA molecules corresponding to each second allele of the locus. It is calculated. For example, the methylation density of the CpG site corresponding to the locus can be determined.

  At block 950, the first methylation profile of the first tissue is generated from the methylation density of the first locus. The first methylation property may correspond to a particular site, eg a CpG site. The methylation profile can be all loci with fetal-specific alleles, or just a few of those loci.

IV. Use of the Differences between Plasma and Blood Methylome In the above, it was shown that plasma-derived fetal-specific leads correlate with placental methylome. Because the maternal component of the maternal plasma methylome is mainly provided by blood cells, the differences between plasma and blood methylomes are used to determine the placental methylome of all loci, not just the location of the fetal-specific alleles. be able to. The difference between plasma and blood methylomes can also be used to determine the tumor methylome.

A. Method FIG. 10 is a flow chart illustrating a method 1000 of determining a first methylation profile from a biological sample of an organism, according to an embodiment of the invention. A biological sample (eg, plasma) includes cell-free DNA containing a mixture of cell-free DNA from a first tissue and a second tissue. The first methylation profile corresponds to the methylation profile of the first tissue (eg, fetal or tumor tissue). Method 1200 may allow inference of differentially methylated regions from maternal plasma.

  At block 1010, a biological sample is received. The biological sample may simply be received at the device (eg, a sequencing device). The biological sample may be in a form obtained from the organism or in a processed form, for example, the sample may be plasma extracted from a blood sample.

  At block 1020, a second methylation profile corresponding to the DNA of the second tissue is obtained. The secondary methylation profile has been previously determined and can be read from memory. The second methylation profile can be determined from a second tissue (eg, a different sample containing only or predominantly cells of the second tissue). The secondary methylation profile can correspond to the cellular methylation profile and can be obtained from cellular DNA. As another example, the second characteristic is that before pregnancy, or because the plasma methylome of a cancer-free non-pregnant person is very similar to that of blood cells, before the development of cancer, It can be determined from the collected plasma sample.

  The second methylation profile may provide the methylation density at each of multiple loci within the organism's genome. The methylation density at a particular locus corresponds to the percentage of methylated second tissue DNA. In one embodiment, the methylation density is the CpG methylation density, where the CpG site associated with the locus is used to determine the methylation density. If there is one site at the locus, the methylation density can be the same as the methylation index. Since the values of the methylated density and the unmethylated density are complementary, the methylated density also corresponds to the unmethylated density.

  In one embodiment, methylation recognition sequencing of cellular DNA from a sample of the organism is performed to obtain a second methylation profile. One example of methylation recognition sequencing involves treating the DNA with sodium bisulfite followed by DNA sequencing. In another example, methylation recognition sequencing determines the methylation status of DNA molecules (eg, N6-methyladenine, 5-methylcytosine and 5-hydroxymethylcytosine) without the use of sodium bisulfite. Using a single molecule sequencing platform that allows direct elucidation without salt conversion (AB Flusberg et al. 2010 Nat Methods; 7: 461-465; J Shim et al. 2013 Sci Rep; 3: 1389. doi: 10.1038 / srep01389); or immunoprecipitation of methylated cytosine (eg, by using an antibody against methylcytosine or by using a methylated DNA binding protein or peptide) (LG Acevedo et al. 2011 Epigenomics; 3: 93-101), followed by sequencing; or by the use of methylation sensitive restriction enzymes followed by sequencing. In another embodiment, non-sequencing methods such as arrays, digital PCR and mass spectrometry are used.

  In another embodiment, the second tissue methylation density can be previously obtained from a control sample of the subject, or from another subject. A methylation density from another subject can serve as a reference methylation property with a reference methylation density. The reference methylation density can be determined from multiple samples, where the average level of different methylation densities at a locus (or other statistic) can be used as the reference methylation density at that locus.

  At block 1030, cell-free methylation properties are determined from the cell-free DNA of the mixture. The cell-free methylation profile gives the methylation density at each of the multiple loci. Cell-free methylation properties can be determined by obtaining sequence reads from sequencing cell-free DNA and using the sequence reads to obtain methylation information. Cell-free methylation properties can be determined similar to cell methylome.

  At block 1040, the percentage of cell-free DNA from the first tissue in the biological sample is determined. In one embodiment, the first tissue is fetal tissue and the corresponding DNA is fetal DNA. In another embodiment, the first tissue is tumor tissue and the corresponding DNA is tumor DNA. The ratio can be determined in various ways, for example using a fetal-specific or tumor-specific allele. By using copy number, for example, US Patent Application Publication No. 13 / 801,748, filed March 13, 3013, entitled "Mutational Analysis Of Plasma DNA For Cancer Detection," incorporated by reference. Percentages can also be determined as described in.

  At block 1050, a plurality of loci for determining the first methylome are identified. These loci may correspond to each locus used to determine cell-free and secondary methylation properties. Therefore, multiple loci can be matched. It is possible that more loci can be used to determine cell-free and secondary methylation properties.

  In some embodiments, hypermethylated or hypomethylated loci in the secondary methylation profile can be identified using, for example, maternal blood cells. CpG sites with a methylation index greater than or equal to X% (eg, 80%) can be scanned from one end of the chromosome to identify loci that are hypermethylated in maternal blood cells. Next, the next CpG site within the downstream region (eg, within 200 bp downstream) can be searched. The first CpG site and the second CpG site may be grouped if the immediately downstream CpG site had a methylation index greater than or equal to X% (or other specified amount). Grouping can be continued until no other CpG sites are present in the next downstream region; or immediately downstream CpG sites have a methylation index of less than X%. Regions of grouped CpG sites can be reported as hypermethylated in maternal blood cells if the region contains at least 5 immediately adjacent hypermethylated CpG sites. By performing the same analysis, a locus in a hypomethylated state can be searched for in a maternal blood cell for a CpG site having a methylation index of 20% or less. The methylation density of the second methylation property can be calculated for short-listed loci, and the first methylation property of the corresponding locus (eg placental tissue methylation density) can be calculated as For example, it can be used to extrapolate from maternal plasma bisulfite sequencing data.

  At block 1060, the first methylation profile of the first tissue has a differential parameter including a difference between a methylation density of the second methylation profile and a methylation density of the cell-free methylation profile for each of the plurality of loci. It is determined by calculating. Differences are determined as a percentage.

  In one embodiment, the first methylation density (D) of the locus in the first (eg, placenta) tissue is estimated using the formula:

Where mbc is the methylation density of the second methylation profile at the locus (eg, the shortly listed locus determined in maternal blood cell bisulfite sequencing data); mp is maternal plasma bisulfite Denotes the methylation density of the corresponding loci in the salt sequencing data; f is the percentage of cell-free DNA from the first tissue (eg fetal DNA concentration fraction), CN is the copy number at the locus (eg, Higher amplification value or lower number of deletions) compared to normal. The CN may be one if there is no amplification or deletion in the first tissue. In trisomy (ie duplication of regions in the tumor or fetus), the CN is 1.5 (because the increase is 2 to 3 copies) and the monosomy has 0.5. Higher amplification can be increased by a value of 0.5. In this example, D may correspond to the differential parameter.

  At block 1070, the first methylation density is transformed to obtain a corrected first methylation density of the first tissue. The transformation may give a fixed difference between the differential parameters and the actual methylation properties of the first tissue. For example, the values can differ by a fixed constant or by a slope. The transformation can be linear or non-linear.

  In one embodiment, the distribution of estimates D was found to be lower than the actual methylation level of placental tissue. For example, the estimates can be linearly transformed using data obtained from CG islands, which are genomic segments in which CpG sites are over-represented. The genomic location of CG islands used in this study was obtained from the UCSC Genome Browser database (NCBI build 36 / hg18) (PA Fujita et al. 2011 Nucleic Acids Res; 39: D876-882). For example, a CG island can be defined as a genomic segment with a GC content of 50% or greater, a genome length of greater than 200 bp and a measured / predicted CpG number ratio of greater than 0.6 (M Gardiner-Garden et al 1987 J Mol Biol; 196: 261-282).

In some implementations, at least 4 CpG sites and CG islands with an average reading depth of 5 or more per CpG site in the sequenced sample may be included to obtain a linear transformation. After determining the linear relationship between the methylation density of CG islands in CVS or late placenta and the estimated value D, the predicted value was determined using the following formula:
First-term forecast = D x 1.6 + 0.2
Third-term forecast = D x 1.2 + 0.05

B. Fetal Example As described above, method 1000 can be used to estimate placental methylation status from maternal plasma. Circulating DNA in plasma is mainly derived from hematopoietic cells. Note that the proportion of cell-free DNA given by another unknown internal organ is unknown. Furthermore, placenta-derived cell-free DNA accounts for approximately 5-40% of the total DNA in maternal plasma, with an average value of approximately 15%. Therefore, it can be assumed that the methylation level in maternal plasma is equal to the existing background methylation + placental contribution during pregnancy, as described above.

  Maternal plasma methylation level MP can be determined using the formula:

Where BKG is the background DNA methylation level in plasma obtained from blood cells and internal organs, PLN is the placenta methylation level, and f is the fetal DNA concentration fraction in maternal plasma.

  In one embodiment, placental methylation levels can be theoretically estimated by the formula:

Equations (1) and (2) are equal if CN equals 1, D equals PLN and BKG equals mbc. In another embodiment, the fetal DNA concentration fraction can be hypothesized or set to a particular value (eg, as part of the hypothesis that a minimum value f exists).

  Maternal blood methylation levels were considered to represent background methylation of maternal plasma. The putative approach was further sought by focusing on defined regions with clinical relevance, such as CG islands in the human genome, in addition to hypermethylated or hypomethylated loci in maternal blood cells.

  Mean methylation densities of a total of 27,458 CG islands (NCBI Build36 / hg18) on the autosomal and X chromosomes were obtained from maternal plasma and placental sequencing data. Only CG islands with 10 or more coated CpG sites and an average reading depth of 5 or more per coated CpG site were selected in all analytical samples including placenta, maternal blood and maternal plasma. As a result, 26,698 CG islands (97.2%) remained valid and their methylation levels were estimated using plasma methylation data and fetal DNA concentration fractions according to the above equation.

  It was noted that the distribution of estimated PLN values was lower than the actual methylation level of CG islands in placental tissue. Thus, in one embodiment, the estimated PLN value, or simply the estimated value (D), was used as an arbitrary unit to estimate the methylation level of CG islands in the placenta. After transformation, the linearly estimated values and their distributions were more similar to the actual dataset. The transformed estimate was named the methylation predictive value (MPV) and was then used to predict the methylation level of the locus in the placenta.

  In this example, CG islands were classified into three categories based on their methylation densities in the placenta: low (less than 0.4), medium (greater than 0.4 to less than 0.8) and high (0). .8 or more). The estimation formula was used to calculate the MPV of the same series of CG islands, and then the values were used to classify the CG islands into three categories with the same cutoff. By comparing the actual and estimated datasets, it was found that 75.1% of the shortly listed CG islands could exactly match the same category in their MPV-tissue data. Approximately 22% of CG islands were assigned to groups with one level of difference (high vs. medium or medium vs. low), and less than 3% were completely misclassified (high vs. low) (Figure 12A). Also, the overall classification performance was determined: 86.1%, 31.4% and 68 with methylation densities greater than 0.4, greater than 0.4 and less than 0.8 and greater than 0.8 in placenta. CG islands of 0.8% were accurately estimated as "low", "medium" and "high" (Fig. 12B).

  11A and 11B show graphs of performing a prediction algorithm using maternal plasma data and fetal DNA concentration fractions, according to embodiments of the invention. FIG. 11A shows MPV corrected classification (estimated category exactly matches the actual dataset); one level difference (estimated category differs from the actual dataset by one level); misclassification (estimated category differs from the actual dataset). 1 is a graph 1100 showing the accuracy of CG island classification using (opposite). FIG. 11B is a graph 1150 showing the percentage of CG islands classified into each estimation category.

  If maternal background methylation is low in each genomic region, the presence of highly methylated placenta-derived DNA in the circulating blood increases the overall plasma methylation level to a degree dependent on the fetal DNA concentration fraction. increase. Significant changes can be observed if the fetal DNA released is fully methylated. Conversely, when maternal background methylation is high, the degree of change in plasma methylation levels becomes more significant when hypomethylated fetal DNA is released. Therefore, the estimation scheme is designed for loci whose methylation levels are known to differ between maternal background and placenta, especially for hypermethylated loci and hypomethylation markers in placenta. , Can be more practical.

  FIG. 12A is a table 1200 showing details of 15 selected genomic loci for methylation prediction, according to an embodiment of the invention. To confirm the approach, 15 previously studied differentially methylated genomic loci were selected. Methylation levels in selected regions were estimated and compared to the 15 previously studied differentially methylated loci (RWK Chiu et al. 2007 Am J Pathol; 170: 941-950; SSC Chim. et al. 2008 Clin Chem; 54: 500-511; SSC Chim et al. 2005 Proc Natl Acad Sci USA; 102: 14753-14758; DWY Tsui et al. 2010 PLoS One; 5: e15069).

  FIG. 12B is a graph 1250 showing estimated categories of 15 selected genomic loci and their corresponding methylation levels in placenta. The estimated methylation categories are low, 0.4 or less; medium, greater than 0.4 and less than 0.8; high, 0.8 or more. Table 1200 and graph 1300 show that their methylation levels in placenta can be accurately estimated with some exceptions (RASSF1A, CGI009, CGI137 and VAPA). Of these four markers, only CGI009 showed a significant discrepancy with the actual data set. Others were only slightly misclassified.

  In Table 1200, “1” indicates an estimated value (D) calculated by the following formula,

In the formula, f is the fetal DNA concentration fraction. The label "2" refers to the predicted methylation value (MPV), which refers to the linearly transformed estimate using the formula: MPV = D × 1.6 + 0.25. The label "3" refers to the classification cutoff for the estimate: low, 0.4 or less; medium,> 0.4 to <0.8; high, 0.8 or more. Label "4" refers to the classification cutoff for the actual placenta data set: low, 0.4 or less; medium, greater than 0.4 and less than 0.8; high, 0.8 or greater. The label "5" indicates that the placental status refers to the methylation status of the placenta compared to the methylation status of maternal blood cells.

C. Calculation of Fetal DNA Concentration Fraction In one embodiment, the proportion of fetal DNA derived from the first tissue can be the Y chromosome of the male fetus. The proportion of Y-chromosome sequences in maternal plasma samples (Y-chromosome rate) was a hybrid of the number of Y-chromosome leads from boy fetuses and maternal (female) reads misaligned with the Y chromosome (RWK Chiu et al. 2011). BMJ; 342: c7401). Therefore, the relationship between the Y chromosome rate in the sample and the fetal DNA concentration fraction (f) can be given by:

In the formula,% chrY _male (Y chromosome rate (male)) refers to the percentage of reads aligned with the Y chromosome in a plasma sample containing 100% male DNA;% chrY _female (Y chromosome rate (female)). Refers to the percentage of reads aligned to the Y chromosome in a plasma sample containing 100% female DNA.

Y-chromosome percentage can be determined from reads aligned to the Y-chromosome without mismatches in samples obtained from women who were pregnant with male fetuses, eg, the reads from samples converted by bisulfite. Is what you get. The% chrY _male value can be obtained from bisulfite sequencing of two adult male plasma samples. The% chrY _female value can be obtained from bisulfite sequencing of two non-pregnant adult female plasma samples.

  In another embodiment, fetal DNA rates can be determined from autosomal fetal-specific alleles. As another example, epigenetic markers may be used to determine fetal DNA rates. Other methods of determining fetal DNA rate may be used.

D. Methods for determining copy number using methylation The placental genome is more hypomethylated than the maternal genome. As mentioned above, the methylation of plasma in pregnant women depends on the concentration fraction of placental-derived fetal DNA in maternal plasma. Therefore, it is possible to detect differences in the contribution of fetal tissue to maternal plasma through analysis of the methylation density of chromosomal regions. For example, in a pregnant woman harboring a trisomy fetus (eg, a fetus carrying trisomy 21 or trisomy 18 or trisomy 13), the fetus is further transferred from the trisomic chromosome to maternal plasma when compared to the disomal chromosome. Give a quantity of DNA. In this situation, the plasma methylation density of trisomic chromosomes (or any chromosomal region with amplification) will be lower than that of disomic chromosomes. The degree of difference can be predicted by mathematical calculations by taking into account the fetal DNA concentration fraction in the plasma sample. The higher the fetal DNA concentration fraction in the plasma sample, the greater the difference in the methylation density between the trisomic and disomic chromosomes. For regions with deletions, the methylation density will be higher.

  One example of a deletion is Turner's syndrome when a female fetus has only one copy of the X chromosome. In this situation, in pregnant women carrying a fetus with Turner syndrome, the methylation density of the X chromosome in plasma DNA is higher than in the same pregnant woman carrying a female fetus with a normal number of X chromosomes. Become. In one embodiment of this strategy, maternal plasma may first be analyzed for the presence or absence of Y chromosome sequences (eg, using MPS or PCR based techniques). If the Y chromosome sequence is present, the fetus can be classified as male and subsequent analysis is not necessary. On the other hand, if the Y chromosome sequence is not present in maternal plasma, the fetus can be classified as female. In this situation, the methylation density of the X chromosome in maternal plasma can then be analyzed. A higher than normal X chromosome methylation density indicates that the fetus is at increased risk of having Turner syndrome. This approach can be applied to other sex chromosome aneuploidies. For example, in a fetus with XYY, the methylation density of the Y chromosome in maternal plasma will be lower than that in a normal XY fetus with similar levels of fetal DNA in maternal plasma. As another example, in a fetus with Kleinfelter's syndrome (XXY), the Y-chromosome sequences are present in maternal plasma, but the methylation density of the X-chromosome in maternal plasma shows similar levels in fetuses in maternal plasma. It will be lower than that of a normal XY fetus with DNA.

From the above consideration, the plasma methylation density (MP _Non-aneu ) of the _disomic chromosome is

Where BKG is the background DNA methylation level in plasma from blood cells and internal organs, PLN is the placenta methylation level, and f is the fetal DNA concentration in maternal plasma. It is a fraction.

Plasma methylation density of _trisomic chromosomes (MP _Aneu ( _{unequal number} )) is

Where 1.5 corresponds to the copy number CN and the addition of another chromosome results in a 50% increase. The difference between the _trisomic and disomy chromosomes (MP _Diff ) is:

  In one embodiment, comparing the methylation density of potentially aneuploid chromosomes (or chromosomal regions) to the overall methylation density of one or more other putative non-aneuploid chromosomes or genomes is It can be used to efficiently normalize fetal DNA concentration in a sample. The comparison may be by calculating a parameter (eg, including a ratio or difference) between the methylation densities of the two regions to obtain a normalized methylation density. The comparison can eliminate the resulting dependence of methylation levels (eg, determined as a parameter from the two methylation densities).

  If the methylation density of a potentially aneuploid chromosome is not normalized to the methylation density of one or more other chromosomes, or other parameters that reflect the concentration fraction of fetal DNA, the concentration Fraction is the main factor affecting the methylation density in plasma. For example, the plasma methylation density of chromosome 21 of a pregnant woman who carries a trisomy 21 fetus having a fetal DNA concentration fraction of 10% is the same as that of a pregnant woman who carries an euploid fetus. The rate is 15%, but the normalized methylation densities show differences.

In another embodiment, the methylation density of potentially aneuploid chromosomes can be normalized to the fetal DNA concentration fraction.
For example, the following equation can be applied to normalize the methylation density,

_Where MP _Normalized is the methylation density normalized using the fetal DNA concentration fraction in plasma, MP _{non-normalized} is the measured methylation density, and BKG is Background methylation density obtained from maternal blood cells or tissues, PLN is the methylation density in placental tissue, and f is the fetal DNA concentration fraction. BKG and PLN methylation densities can be based on previously established baseline values from maternal blood cells and placental tissue obtained from normal pregnant women. Various genetic and epigenetic techniques have been used to measure the proportion of sequence reads from the Y chromosome in plasma samples, for example, by massively parallel sequencing or PCR with non-bisulfite converted DNA. The fetal DNA concentration fraction of can be determined.

In some implementations, the normalized methylation density of potentially aneuploid chromosomes can be compared to a reference group consisting of pregnant women bearing euploid fetuses. The mean and SD of the normalized methylation densities of the reference group can be determined. The normalized methylation density of the tested cases can then be expressed as a Z value that represents the number of SD from the mean of the reference group according to the formula:

_Where MP _Normalized is the normalized methylation density of the tested cases, Mean (mean) is the averaged normalized methylation density of the reference cases, and SD is the normalized methylation of the reference cases. It is the standard deviation of the methylation density. A cutoff (eg, a Z value of less than -3) can be used to classify whether chromosomes are significantly hypomethylated, thereby determining the aneuploid state of the sample.

In another embodiment, MP _Diff can be used as the normalized methylation density. In such an embodiment, PLN may be estimated using, for example, method 1000. In some implementations, the reference methylation density (which can be normalized with f) can be determined from the methylation level of the non-aneuploid region. For example, the Mean can be determined from one or more chromosomal regions of the same sample. The cutoff is determined as a function of f, or simply set to a level sufficient to have a minimum concentration present.

  Thus, comparing the methylation level of a region to the cutoff can be achieved in various ways. The comparison can include normalization (eg, as described above), and the normalization can be performed equally on methylation levels or cutoff values, depending on the way those values were defined. Thus, whether the determined methylation level of a region is statistically different from the reference level (determined from the same or other samples) can be determined in various ways.

  The above analysis can be applied to the analysis of chromosomal regions, which can include whole chromosomes or parts of chromosomes (eg, contiguous or discrete small regions of a chromosome). In one embodiment, potentially aneuploid chromosomes may be divided into bins. The bins can be of the same or different sizes. The methylation density of each bin can be normalized to the concentration fraction of the sample, or to the methylation density of one or more putative non-aneuploid chromosomes or the overall methylation density of the genome . The normalized methylation density of each bin can then be compared to a reference group to determine if it is significantly hypomethylated. The percentage of bins that are significantly hypomethylated can then be determined. Cutoffs (eg,> 5%, 10%, 15%, 20%, or 30% of bins are significantly hypomethylated) can be used to classify case aneuploidal status.

  When testing for amplification or deletion, the methylation density can be compared to a reference methylation density, which can be specific to the particular region tested. Each region may have a different reference methylation density, as the methylation may differ from region to region, and in particular depending on the size of the region (eg smaller regions show greater variation).

  As described above, by using one or more pregnant women who carry euploid fetuses, respectively, we define the normal range of methylation density in the target region, or the difference in methylation density between two chromosomal regions. can do. The normal range of PLN can also be determined (eg, by direct measurement or as an estimate by method 1000). In other embodiments, a ratio between the two methylation densities can be used, for example, the methylation densities of potentially aneuploid and non-aneuploid chromosomes can be analyzed instead of their differences. Can be used. This methylation analysis approach includes a sequence read counting approach (RWK Chiu et al. 2008 Proc Natl Acad Sci USA; 105: 20458-20463) and an approach including size analysis of plasma DNA (US Patent Application Publication No. 2011/0276277). In combination with, aneuploidy can be determined or confirmed. The sequence read counting approach used in combination with methylation analysis was based on a random sequence (RWK Chiu et al. 2008 Proc Natl Acad Sci USA; 105: 20458-20463; DW Bianchi DW et al. 2012 Obstet Gynecol 119: 890-901. ) Or the target sequence (AB Sparks et al. 2012 Am J Obstet Gynecol 206: 319.e1-9; B Zimmermann et al. 2012 Prenat Diagn 32: 1233-1241; GJ Liao et al. 2012 PLoS One; 7: e38154) Can be done using.

  The use of BKG may explain the difference in background between samples. For example, one woman may have a different BKG methylation level than another, but the difference between BKG and PLN can be used across samples in such situations. The cutoffs of different chromosomal regions can be different, for example, if the methylation density of one region of the genome differs from another region of the genome.

  This approach can be generalized to detect any chromosomal abnormality, including deletions and amplifications in the fetal genome. Furthermore, the resolution of this analysis can be adjusted to the desired level, for example the genome can be divided into 10Mb, 5Mb, 2Mb, 1Mb, 500kb, 100kb bins. Thus, the technology can also be used to detect subchromosomal duplications or deletions. Therefore, the present technology makes it possible to non-invasively obtain the molecular karyotype of the prenatal fetus. When used in this way, the technology is a non-invasive prenatal test based on molecular enumeration (A Srinivasan et al. 2013 Am J Hum Genet; 92: 167-176; SCY Yu et al. 2013 PLoS One. 8: e60968). In other embodiments, the bin sizes may not be the same. For example, the size of the bins can be adjusted so that each bin contains the same number of CpG dinucleotides. In this case, the physical size of the bin will be different.

  The above equation can be rewritten as follows to apply to various types of chromosomal abnormalities.

In the formula, CN represents the number of copy number changes in the damaged area. CN is equal to 1 in an increase in one copy of the chromosome, is equal to 2 in an increase in two copies of the chromosome, and is equal to -1 in a decrease of one of the two homologous chromosomes (e.g. Detection of fetal Turner syndrome in which one of the X chromosomes has been lost). It is not necessary to change this formula when changing the size of the bin. However, the lower the number of CpG dinucleotides (or other nucleotide combinations that show differential methylation between fetal and maternal DNA), the more they are in the smaller bins, which leads to a higher methylation density. Due to the increased stochastic variability in the measurements, sensitivity and specificity can be reduced when smaller bin sizes are used. In one embodiment, the number of reads required can be determined by analyzing the coefficient of variation of methylation density and the desired sensitivity level.

  To show that this approach is feasible, plasma samples obtained from 9 pregnant women were analyzed. Of the 5 pregnant women, each had euploid fetuses, and the other four had trisomy 21 (T21) fetuses each. From the 5 euploid pregnant women, 3 were randomly selected to form a reference group. The remaining two euploid pregnancies (Eu1 and Eu2) and four T21 cases (T21-1, T21-2, T21-3 and T21-4) were tested for potential T21 status. It was analyzed using this approach. Plasma DNA was converted with bisulfite and sequenced using the Illumina HiSeq2000 platform. In one embodiment, the methylation density of individual chromosomes was calculated. The difference in methylation density between the mean values of chromosome 21 and the other 21 autosomes was then determined to give the normalized methylation density (Table 1). The average value and SD of the reference group were used to calculate the Z value of the 6 test examples.

  In another embodiment, the genome was divided into 1 Mb bins and the methylation density of each 1 Mb bin was determined. The methylation densities of all bins on potentially aneuploid chromosomes can be normalized using the median methylation densities of all bins located on putative non-aneuploid chromosomes. In one implementation, the difference in methylation density from the median of non-aneuploid bins can be calculated for each bin. The Z value of these values can be calculated using the average value and standard deviation value of the reference group. The percentage of bins exhibiting hypomethylation (Table 2) can be determined and compared to the cutoff rate.

  This DNA methylation-based approach for detecting fetal chromosomes or abnormalities within chromosomes has been sequenced (RWK Chiu et al. 2008 Proc Natl Acad Sci USA; 105: 20458-20463) or digital PCR (YMD Lo et al. al. 2007 Proc Natl Acad Sci USA; 104: 13116-13121) or the like, or can be used in combination with an approach based on size measurement of DNA molecules (US Patent Application Publication No. 2011/0276277). Such a combination (eg, DNA methylation + molecule count, or DNA methylation + size measurement, or DNA methylation + molecule count + size measurement) may be advantageous in clinical settings, such as increased sensitivity and / or specificity. Will have a synergistic effect. For example, the number of DNA molecules that need to be analyzed, for example by sequencing, can be reduced without adversely affecting diagnostic accuracy. This feature makes it possible to carry out such tests more economically. As another example, for a given number of DNA molecules analyzed, the combined approach allows detection of fetal chromosomes or intrachromosomal abnormalities in lower concentrations of fetal DNA.

  FIG. 13 is a flow chart of a method 1300 for detecting a chromosomal abnormality from a biological sample of an organism. The biological sample includes cell-free DNA containing a mixture of cell-free DNA from the first tissue and the second tissue. The first tissue may be obtained from a fetus or tumor and the second tissue may be obtained from a pregnant female or patient.

  At block 1310, a plurality of DNA molecules from the biological sample are analyzed. Analyzing a DNA molecule can include determining the location of the DNA molecule in the organism's genome and determining whether the DNA molecule is methylated at one or more sites. Since the analysis can be performed by obtaining a sequence read from methylation recognition sequencing, the analysis can only be performed on data previously obtained from that DNA. In other embodiments, the analysis may include actual sequencing or other valid steps to obtain data.

  Positioning can include mapping the DNA molecule to each part of the human genome, eg, to a specific region (eg, via a sequence read). In some implementations, a lead may be ignored if it is not located in a lead target area.

  At block 1320, for each of the plurality of sites, the number of each DNA molecule that is methylated at the site is determined. In one embodiment, the site is a CpG site, and may only be one particular CpG site selected using one or more of the criteria described herein. The number of DNAs that are methylated can be used to determine the number that is unmethylated if normalization was performed using the total number of DNA molecules analyzed at a particular site (eg, total number of sequence reads). equal.

  At block 1330, the first methylation level of the first chromosomal region is calculated based on the respective number of DNA molecules that are methylated at sites within the first chromosomal region. The first chromosomal region can be of any size (eg, the size described above). The methylation level can explain, for example, the total number of DNA molecules aligned to the first chromosomal region as part of the normalization procedure.

  The first chromosomal region can be of any size (eg, an entire chromosome) and can be composed of discrete subregions (ie, the subregions are separated from each other). The methylation level of each subregion can be determined and combined (eg, as a mean or median) to determine the methylation level of the first chromosomal region.

  At block 1340, the first methylation level is compared to the cutoff value. The cutoff value may be the reference methylation level or it may be associated with the reference methylation level (eg, a particular distance from normal levels). Cutoff values are known not to be associated with other pregnant female subjects who carry fetuses without chromosomal abnormalities in the first chromosome region, from samples of individuals without cancer, or with aneuploidy. It can be determined from the locus of the organism (ie, the disomy region) in which it is maintained.

  In one embodiment, the cutoff value may be defined as having a difference from a reference methylation level of the formula:

Where BKG is the female background (or mean or median from other subjects), f is the concentration fraction of cell-free DNA from the first tissue, and CN is the copy number tested. is there. CN is an example of a scaling factor corresponding to the type of abnormality (deletion or duplication). A CN = 1 cutoff can be used first to check all amplifications, and then a further cutoff can be used to determine the extent of amplification. The cut-off value can be based on the concentration fraction of cell-free DNA from the first tissue to determine the predictive level of methylation at the locus (eg, in the absence of copy number abnormalities).

  At block 1350, a classification of abnormalities in the first chromosomal region is determined based on the comparison. Statistically significant differences in levels may indicate an increased risk for fetuses with chromosomal abnormalities. In various embodiments, the chromosomal abnormality can be trisomy 21, trisomy 18, trisomy 13, triner Turner syndrome, or Kleinfelter syndrome. Other examples are intrachromosomal deletions, intrachromosomal duplications, or DiGeorge syndrome.

V. Marker Determination As mentioned above, certain parts of the fetal genome are methylated differently from the maternal genome. These differences can be common throughout pregnancy. Regions of different methylation can be used to identify fetal DNA fragments.

A. Methods for Determining DMR from Placental and Maternal Tissue Placenta has a tissue-specific methylation signature. A fetal-specific DNA methylation marker was developed for maternal plasma detection and for non-invasive prenatal diagnostic applications based on the loci differentially methylated between placental tissue and maternal blood cells. (SSC Chim et al. 2008 Clin Chem; 54: 500-511; EA Papageorgiou et al 2009 Am J Pathol; 174: 1609-1618; and T Chu et al. 2011 PLoS One; 6: e14723). Embodiments are provided for genome-wide searching for such differentially methylated regions (DMRs).

  FIG. 14 is a flow chart of a method 1400 for identifying methylation markers by comparing placental methylation properties to maternal methylation properties (eg, determined from blood cells) according to embodiments of the present invention. . Method 1400 may be used to determine tumor markers by comparing tumor methylation profiles to methylation profiles corresponding to healthy tissue.

  At block 1410, a placenta methylome and a blood methylome are obtained. Placental methylome can be determined from placental samples (eg, CVS or late pregnancy placenta). It is to be understood that the methylome may contain the methylation density of only part of the genome.

  At block 1420, a region containing a particular number of sites (eg, 5 CpG sites) was identified and a sufficient number of reads were obtained for that. In one embodiment, identification was initiated at one end of each chromosome to locate the first 500 bp region containing at least 5 suitable CpG sites. A CpG site may be considered suitable if it is covered by at least 5 sequence reads.

  At block 1430, a placental methylation index and a blood methylation index are calculated for each site. For example, the methylation index was calculated individually for all appropriate CpG sites within each 500 bp region.

  At block 1440, the methylation index was compared between maternal blood cells and placental samples to determine if the series of indices differed from each other. For example, the methylation index was compared between maternal blood cells and CVS or late pregnancy placenta using, for example, the Mann-Whitney test. For example, P values of 0.01 or less were considered to be statistically significantly different, but other values may be used if lower numbers reduce false positive areas.

  In one embodiment, the 500 bp region was moved 100 bp downstream if the number of suitable CpG sites was less than 5, or if the Mann-Whitney test was not significant. The region was allowed to move downstream until the Mann-Whitney test was significant for the 500 bp region. Next, the next 500 bp region was considered. If the next region was found to show statistical significance by the Mann-Whitney test, it was added to the current region unless the combined flanking region was less than 1,000 bp.

  At block 1450, adjacent regions that were statistically significantly different (eg, by the Mann-Whitney test) may be integrated. Note that there is a difference between the methylation indices of the two samples. In one embodiment, adjacent regions are integrated if they are within a certain distance of each other (eg, 1,000 bp) and they show similar methylation properties. In certain implementations, the similarity of methylation properties between adjacent regions can be defined using any of the following: (1) For maternal blood cells, the same trend in placental tissue (eg, both regions show More methylated in placental tissue); (2) with less than 10% difference in methylation density in adjacent regions in placental tissue; and (3) less than 10% methylated in adjacent regions in maternal blood cells. Have a difference in density.

  At block 1460, the methylation density of blood methylome from maternal blood cell DNA and placental samples (eg, CVS or late pregnancy placental tissue) in the region is calculated. Methylation density can be determined as described herein.

At block 1470, putative DMRs at which the total placental and blood methylation densities are statistically significantly different for all sites in the region are determined. In one embodiment, all suitable CpG sites within the integrated region undergo a χ ² test. The χ ² test showed that the number of methylated cytosines as a percentage of methylated and unmethylated cytosines between all relevant CpG sites within the integrated region was statistically significant between maternal blood cells and placental tissues. It was evaluated whether it was different. In some implementations, a P value of 0.01 or less may be considered statistically significantly different in the χ ² test. Integrated segments that were shown to be significant by the χ ² test were considered putative DMRs.

  At block 1480, loci in which the methylation density of maternal blood cell DNA was above the high cutoff or below the low cutoff were identified. In one embodiment, loci were identified in which the maternal blood cell DNA methylation density was 20% or less or 80% or more. In other embodiments, body fluids other than maternal blood may be used, including but not limited to saliva, uterine or cervical lavage fluid from female reproductive organs, tears, sweat, saliva, and urine.

  Important for successful development of fetal-specific DNA methylation markers in maternal plasma is that the methylation status of maternal blood cells is as highly or as methylated as possible. Can be not. This may reduce (eg, minimize) the likelihood of having a maternal DNA molecule that interferes with the analysis of placental-derived fetal DNA molecules that exhibit opposite methylation properties. Therefore, in one embodiment, candidate DMRs were selected by further screening. Candidate hypomethylation loci were those that showed a methylation density of 20% or less in maternal blood cells and a methylation density of at least 20% higher in placental tissue. Candidate hypermethylation loci were those that showed a methylation density of 80% or higher in maternal blood cells and a methylation density of at least 20% lower in placental tissues. Other percentages may be used.

  At block 1490, DMRs were then identified among some loci where placental methylation density was significantly different from blood methylation density by comparing the difference to a threshold. In one embodiment, the threshold is 20%, so the methylation density was at least 20% different from the methylation density of maternal blood cells. Thus, the difference between placental and blood methylation densities at each identified locus can be calculated. The difference can be a simple subtraction. In other embodiments, scaling factors and other functions can be used to determine the difference (eg, the difference can be the result of a function applied to a simple subtraction).

  In one implementation, 11,729 hypermethylated loci and 239,747 hypomethylated loci were identified from first stage placental samples using this method. The top 100 hypermethylated loci are listed in Appendix S, Table S2A. The top 100 hypomethylated loci are listed in Table S2B in the Appendix. Tables S2A and S2B list chromosomes, start and stop positions, size of region, methylation density in maternal blood, methylation density in placental samples, P-values (all very small), and methylation differences. ing. The position corresponds to the reference genome hg18, which is located in hgdownload. soe. ucsc. It can be found at edu / goldenPath / hg18 / chromosomes.

  11,920 hypermethylated loci and 204,768 hypomethylated loci were identified from third stage placental samples. The top 100 hypermethylated loci of the third stage are listed in Table S2C and the top 100 hypomethylated loci are listed in Table S2D. The 33 loci previously reported to be differentially methylated between maternal blood cells and first stage placental tissue were used to validate the enumeration of first stage candidates. 79% of the 33 loci were identified as DMRs using this algorithm.

  FIG. 15A is a table 1500 showing the performance of the DMR identification algorithm using first trimester data with reference to 33 previously reported trimester markers. In the table, "a" refers to loci 1 to 15 previously (RWK Chiu et al. 2007 Am J Pathol; 170: 941-950 and SSC Chim et al. 2008 Clin Chem; 54: 500-511). What was reported; loci 16-23 were previously reported in (KC Yuen, thesis 2007, The Chinese University of Hong Kong, Hong Kong); and loci 24-33 were (EA Papageorgiou et al. 2009). Am J Pathol; 174: 1609-1618). "B" indicates that these data were obtained from the above publications. “C” indicates that the methylation densities of maternal blood cells and chorionic villus specimens and their differences were observed from the genomic coordinate-based sequencing data produced in this study but provided by the original study. It shows that. "D" is the bisulfite sequence without reference to the above publications by Chiu et al (2007), Chim et al (2008), Yuen (2007) and Papageorgiou et al (2009) for data on loci. 6 illustrates that it has been identified using an embodiment of method 1400 for decision data. The full length of the locus contained previously reported genomic regions, but generally spanned larger regions. “E” is a true positive (TP) based on the need for the candidate DMR to observe more than 0.20 differences between the methylation densities of the corresponding genomic coordinates of DMR in maternal blood cells and chorionic villus samples. Or it has shown that it was classified into the false positive (FN).

  FIG. 15B is a table 1550 showing the performance of the DMR identification algorithm compared to placental samples obtained at parturition using the third trimester data. "A" indicates that the same listing of 33 loci as described in Figure 17A was used. "B" indicates that 33 loci were previously identified from early pregnancy samples, so they may not be applicable to trimester data. Therefore, the bisulfite sequencing data generated in this study on late pregnancy placental tissue based on the genomic coordinates provided by the original study was reviewed. Greater than 0.20 differences in methylation densities between maternal blood cells and late pregnancy placental tissue were used to determine whether the locus was indeed a true DMR in third trimester. "C" is for bisulfite sequencing without reference to data published by Chiu et al (2007), Chim et al (2008), Yuen (2007) and Papageorgiou et al (2009) for loci data. It is shown that it was identified using the method 1400 for the data. The full length of the locus contained previously reported genomic regions, but generally spanned larger regions. “D” indicates that a candidate DMR containing a locus suitable for differential methylation in the third trimester is between 0. 0 between the methylation densities of the corresponding genomic coordinates of the DMR in maternal blood cells and late pregnancy placental tissue. It shows that they were classified as true positive (TP) or false positive (FN) based on the need to observe more than 20 differences. For loci that were not suitable for differential methylation in the third phase, the absence of those in the DMR enumeration or the presence of DMRs containing the locus but showing a methylation difference of less than 0.20, It was considered a true negative (TN) DMR.

B. DMR obtained from maternal plasma sequencing data
If the fetal DNA concentration fraction of the sample was also pre-existing, placental tissue DMRs could be directly identified from the bisulfite sequencing data of maternal plasma DNA. Since the placenta may be the major source of fetal DNA in maternal plasma (SSC Chim et al. 2005 Proc Natl Acad Sci USA 102, 14753-14758), methyl methylation of fetal-specific DNA in maternal plasma was used in this study. It has been shown that the maturation status correlates with placental methylome.

  Thus, instead of using a placental sample, an embodiment of method 1400 may be performed using a plasma methylome to determine a putative placental methylome. Therefore, method 1000 and method 1400 can be combined to determine a DMR. Using method 1000, predictive values for placental methylation properties can be determined and used in method 1400. In this analysis, the Examples also focus on loci that were methylated below 20% or above 80% in maternal blood cells.

  In some implementations, to estimate hypermethylated loci in placental tissue as compared to maternal blood cells, less than 20% methylation in maternal blood cells, and at least 50 between blood cell methylation density and predictive value. Loci showing a methylation of 60% or more by the predicted value with a difference of% were selected. To estimate loci of hypomethylation status in placental tissue compared to maternal blood cells, methylation of 80% or more in maternal blood cells and at least 50% difference between blood cell methylation density and predicted value should be used. Those loci that had less than 40% methylation by the predicted value were selected.

  FIG. 16 is a table 1600 showing the number of loci predicted to be hypermethylated or hypomethylated based on direct analysis of maternal plasma bisulfite sequencing data. “N / A” means not applicable. "A" indicates that the search for hypermethylated loci was initiated from the locus list showing less than 20% methylation density in maternal blood cells. "B" indicates that the search for hypomethylated loci started with the locus list showing methylation densities greater than 80% in maternal blood cells. “C” is used to verify the validity of the first trimester maternal plasma data from bisulfite sequencing data obtained from chorionic villus specimens, and the third trimester maternal plasma data to be verified by late pregnancy placental tissues. It has been used to do.

  As shown in Table 1600, the majority of non-invasively estimated loci displayed predicted methylation patterns in tissues, overlapping with DMRs mined from tissue data and shown in the previous section. . The appendix lists DMRs identified from plasma. Table S3A lists the 100 loci from the top that were estimated to be hypermethylated from the bisulfite sequencing data of first-stage maternal plasma. Table S3B lists the 100 loci from the top that were estimated to be hypomethylated from the bisulfite sequencing data of first stage maternal plasma. Table S3C lists the 100 loci from the top that were estimated to be hypermethylated from the bisulfite sequencing data of the third trimester maternal plasma. The top 100 loci estimated to be hypomethylated from the bisulfite sequencing data of the third stage maternal plasma are listed.

C. Placental Methylome and Fetal Methylome Pregnancy Variation The total percentage of methylated CpG in CVS was 55%, while the total percentage of methylated CpG in late pregnancy placenta was 59% (Table 100 in Figure 1). The lower methylated DMRs could be identified from CVS more than the late placenta, but the number of hypermethylated DMRs was similar in the two tissues. Therefore, it was clear that CVS was more hypomethylated than the late pregnancy placenta. This fertility trend was also evident in maternal plasma data. The proportion of methylated CpG between fetal-specific reads was 47.0% in first trimester maternal plasma, but 53.3% in third trimester maternal plasma. The number of hypermethylated loci identified was similar in the first (1,457 locus) and third stage (1,279 locus) maternal plasma samples, but not in the third stage sample (12,677). There were substantially more hypomethylated loci in the first sample (21,812 locus) than in the locus (Table 1600 in Figure 16).

D. Use of Markers Differentially methylated markers, or DMRs, are useful in some embodiments. The presence of such markers in maternal plasma indicates and confirms the presence of fetal or placental DNA. This confirmation can be used as a quality control for non-invasive prenatal testing. DMR serves as a general fetal DNA marker in maternal plasma and has advantages over markers that rely on genotypic differences between the mother and fetus, such as genetic polymorphism-based markers or Y-chromosome-based markers. obtain. DMR is a common fetal marker useful for any pregnancy. Markers based on genetic polymorphisms are applicable only to some pregnancies where the fetus has inherited the marker from its father and whose mother does not carry the marker in its genome. In addition, fetal DNA concentrations in maternal plasma samples can be measured by quantifying those DMR-derived DNA molecules. By knowing the characteristics of DMR expected for normal pregnancy, pregnancy-related complications, especially placental tissue changes, can be observed by observing deviations of maternal plasma DMR characteristics or methylation characteristics from those expected for normal pregnancy. Pregnancy related complications associated with can be detected. Pregnancy-related complications associated with placental tissue changes include, but are not limited to, fetal chromosomal aneuploidy. Examples include trisomy 21, preeclampsia, intrauterine growth restriction and preterm birth.

E. Kits Using Markers Embodiments may provide compositions and kits for performing the methods described herein and other applicable methods. The kit can be used to perform an assay that analyzes fetal DNA (eg, cell-free fetal DNA in maternal plasma). In one embodiment, the kit may include at least one oligonucleotide useful for specific hybridization with one or more loci identified herein. The kit may also include at least one oligonucleotide useful for specific hybridization with one or more reference loci. In one embodiment, the placental hypermethylation marker is measured. The test locus can be methylated DNA in maternal plasma and the reference locus can be methylated DNA in maternal plasma. A similar kit can be constructed to analyze tumor DNA in plasma.

  In some examples, the kit can include at least two oligonucleotide primers that can be used to amplify at least one section of the target locus (eg, the locus in the appendix) and the reference locus. Instead of, or in addition to, primers, the kit can include labeled probes for detecting DNA fragments corresponding to target and reference loci. In various embodiments, one or more oligonucleotides of the kit corresponds to a locus in the table in the Appendix. Typically, the kit will also provide instructions that guide the user in analyzing the test sample and assessing the physiological or pathological condition in the test subject.

  In various embodiments, a kit is provided for analyzing fetal DNA in a biological sample containing a mixture of fetal DNA and DNA from a female subject having a fetus. The kit may include one or more oligonucleotides that specifically hybridize to at least a partition of the genomic region listed in Tables S2A, S2B, S2C, S2D, S3A, S3B, S3C, and S3D. Thus, any number of oligonucleotides from all of these tables, or (are) only one table can be used. Oligonucleotides can function as primers and can be constructed as a primer pair corresponding to a particular region in the table.

VI. Size and Methylation Density Relationship Plasma DNA molecules are known to exist in the circulating blood in the form of short molecules, with most molecules having a length of approximately 160 bp (YMD Lo et. al. 2010 Sci Transl Med; 2: 61ra91, YW Zheng at al. 2012 Clin Chem; 58: 549-558). Interestingly, our data revealed a link between methylation status and size of plasma DNA molecules. Thus, plasma DNA fragment length is associated with DNA methylation levels. The characteristic size characteristics of plasma DNA molecules suggest that most are associated with mononucleosomes, which may result from enzymatic degradation during apoptosis.

  Circulating DNA is naturally fragmented. Specifically, circulating fetal DNA is shorter than maternally-derived DNA in maternal plasma samples (KCA Chan et al. 2004 Clin Chem; 50: 88-92). Since paired-end alignment allows size analysis of bisulfite treated DNA, it is possible to directly determine whether there is a correlation between the size of plasma DNA molecules and their respective methylation levels. Can be evaluated. This was investigated in maternal plasma and control plasma samples of non-pregnant adult women.

  Paired-end sequencing on both ends of each DNA molecule (including whole molecule sequencing) was used to analyze each sample in this study. The length of the sequenced DNA molecule can be determined by aligning the terminal sequence pair of each DNA molecule with the reference human genome and recording the genomic coordinates of the extreme ends of the sequenced reads. Plasma DNA molecules are naturally fragmented into small molecules, and sequencing libraries of plasma DNA are typically made without any fragmentation step. Therefore, the length estimated by sequencing was representative of the size of the original plasma DNA molecule.

  In a previous study, the size characteristics of fetal and maternal DNA molecules in maternal plasma were determined (YMD Lo et al. 2010 Sci Transl Med; 2: 61ra91). It was shown that plasma DNA molecules had a size similar to mononucleosomes and fetal DNA molecules were shorter than maternal DNA molecules. In this study, the relevance of the methylation status of plasma DNA molecules to their size was determined.

A. Results FIG. 17A is a plot 1700 showing the size distribution of maternal plasma DNA, non-pregnant female control plasma DNA, placental DNA and peripheral blood DNA. In maternal samples and non-pregnant female control plasma, these two bisulfite-treated plasma samples had the most abundant total sequence of 166-167 bp long and 10 bp periodicity of DNA molecules shorter than 143 bp. (YMD Lo et al. 2010 Sci Transl Med; 2: 61ra91).

  FIG. 17B is a plot 1750 of size distribution and methylation profile of maternal plasma, adult female control plasma, placental tissue and adult female control blood. For DNA molecules having the same size and containing at least one CpG site, their average methylation density was calculated. Next, the relationship between the size of DNA molecules and their methylation densities was plotted. Specifically, for sequenced reads covering at least one CpG site, the average methylation density for each fragment length ranging from 50 bp up to 180 bp was determined. Interestingly, the methylation density increased with plasma DNA size and peaked at about 166-167 bp. However, this pattern was not observed in placental DNA samples and control blood DNA samples that were fragmented using the sonication system.

  FIG. 18 shows a plot of methylation density and size of plasma DNA molecules. FIG. 18A is a plot 1800 for first trimester maternal plasma. FIG. 18B is a plot 1850 for third trimester maternal plasma. The data for all sequenced reads covering at least one CpG site is represented by the blue curve 1805. Data for reads that also contained the fetal-specific SNP allele are represented by the red curve 1810. Data for reads that also contained the maternal-specific SNP allele are represented by the green curve 1815.

  Reads that contained the fetal-specific SNP allele were considered to be from a fetal DNA molecule. Reads that contained maternal-specific SNP alleles were considered to be from maternal DNA molecules. In general, DNA molecules with higher methylation densities were longer in size. This trend appeared in both fetal and maternal DNA molecules in both first and third trimesters. The overall size of fetal DNA molecules was shorter than the overall size of maternal DNA molecules reported previously.

  FIG. 19A shows a plot 1900 of methylation density and size of sequenced reads for adult non-pregnant women. Plasma DNA samples obtained from adult non-pregnant women also showed the same association between DNA molecule size and methylation status. On the other hand, the genomic DNA sample was fragmented by an sonication step prior to MPS analysis. As shown in plot 1900, the data obtained from blood cell and placental tissue specimens did not show the same trend. Since cell fragmentation is artificial, size and density are not expected to be related. Since naturally fragmented DNA molecules in plasma show size dependence, it can be assumed that the lower the methylation density, the more likely it is that the molecule will be cleaved into smaller fragments.

  FIG. 19B is a plot 1950 showing the size distribution and methylation profile of fetal-specific and maternal-specific DNA molecules in maternal plasma. Fetal-specific and maternal-specific plasma DNA molecules also showed the same correlation between fragmentation size and methylation level. Both placenta-derived circulating cell-free DNA and maternal circulating cell-free DNA fragment lengths increased with methylation levels. Moreover, their methylation status distributions do not overlap with each other, suggesting that the phenomenon occurs regardless of the original fragment length of the source of circulating DNA molecules.

B. Methods Therefore, the size distribution can be used to estimate the overall methylation rate of plasma samples. This methylation measurement can then be followed during pregnancy, during cancer monitoring, or during processing by continuous measurement of the size distribution of plasma DNA according to the relationships shown in Figures 18A and 18B. Methylation measurements can also be used to look for increased or decreased DNA release from the organ or tissue of interest. For example, DNA methylation signatures specific to particular organs (eg, liver) can be specifically sought and the concentration of these signatures in plasma determined. Since DNA is released into the plasma when cells die, increased levels may mean increased cell death or cytotoxicity in that particular organ or tissue. A decrease in levels from a particular organ may mean that the treatment against injury or pathological processes in that organ is under control.

  FIG. 20 is a flow chart of a method 2000 for estimating the methylation level of DNA in a biological sample of an organism, according to an embodiment of the invention. The methylation level of a particular region of the genome or of the entire genome can be estimated. If a specific region is desired, DNA fragments derived only from that specific region can be used.

  At block 2010, the amount of DNA fragments corresponding to various sizes is measured. For each size of the multiple sizes, the amount of multiple DNA fragments from the biological sample corresponding to the size can be measured. For example, the number of DNA fragments having 140 base length can be measured. The quantity may be recorded as a histogram. In one embodiment, the size of each of a plurality of nucleic acids from a biological sample is measured, and the measurements can be performed individually (eg, by single molecule sequencing of the entire molecule or only the ends of the molecule) or collectively (eg. (For example, by electrophoresis). The size can correspond to a range. Thus, the amount can be the amount of DNA fragments having a size within the specified range. When paired-end sequencing is performed, DNA fragments that are located (aligned) in a particular region (determined by paired sequence reads) can be used to determine the methylation level of that region.

  At block 2020, a first value for the first parameter is calculated based on the amount of DNA fragments in the plurality of sizes. In one aspect, the first parameter provides a statistical measure of size characteristics (eg, histogram) of DNA fragments in a biological sample. Since the parameter is determined from the sizes of a plurality of DNA fragments, it can be referred to as a size parameter.

  The first parameter can be of various forms. One parameter is the ratio of DNA fragments of a particular size or size range to total DNA fragments or to DNA fragments of another size or range. Such a parameter is the number of DNA fragments of a particular size divided by the total number of fragments that can be obtained from a histogram (any data structure giving absolute or relative counts of fragments of a particular size). As another example, a parameter can be the number of fragments of a particular size or within a particular range divided by the number of fragments of another size or range. Division can serve as a normalization function to account for the different numbers of DNA fragments being analyzed for different samples. Normalization can be achieved by analyzing the same number of DNA fragments for each sample, which effectively gives the same result as dividing by the total number of analyzed fragments. Further examples of parameters and for size analysis can be found in US Patent Application Publication No. 13 / 789,553 (incorporated by reference for all purposes).

  At block 2030, the first size value is compared to the reference size value. The reference size value can be calculated from the DNA fragment of the reference sample. To determine the reference size value, the methylation properties of the reference sample can be calculated and quantified, as can the value of the first size parameter. Thus, when the first size value is compared to the reference size value, the methylation level can be determined.

  At block 2040, the methylation level is estimated based on the comparison. In one embodiment, it is determined whether the first value of the first parameter is above or below the reference size value, whereby the methylation level of the current sample is above or below the methylation level relative to the reference size value. Whether or not can be determined. In another embodiment, said comparing is accomplished by inputting a first value into a calibration function. The calibration function can efficiently compare the first value with the calibration value (a series of reference size values) by identifying the point on the curve that corresponds to the first value. The estimated methylation level is then given as the output value of the calibration function.

  Therefore, the size parameter can be calibrated to the methylation level. For example, the methylation level can be measured and associated with a particular size parameter of the sample. The data points from various samples can then be fitted to a calibration function. In some implementations, different calibration functions may be used for different DNA subsets. Therefore, there may be several forms of calibration based on prior knowledge of the association between methylation and size of a particular DNA subset. For example, the calibration of fetal and maternal DNA can be different.

  As indicated above, fetal DNA is smaller due to its lower methylation because the placenta is more hypomethylated compared to maternal blood. Therefore, by using the average size (or other statistic) of the sample fragments, the methylation density can be estimated. This approach can be cost-effective when used clinically, as fragment size can be measured using paired-end sequencing rather than methylation recognition sequencing, which can be technically more complex. is there. This approach eliminates methylation changes associated with the progress of pregnancy or associated with pregnancy-related diseases such as preeclampsia, preterm birth and fetal disorders (eg, disorders caused by chromosomal or genetic abnormalities or intrauterine growth retardation). It can be used for monitoring.

  In another embodiment, this approach can be used for cancer detection and monitoring. For example, successful treatment of cancer will shift the methylation profile in plasma or other body fluids measured using this size-based approach towards that of healthy individuals without cancer. . Conversely, if the cancer is in progress, the methylation profile in plasma or another body fluid will deviate from that of a healthy person without cancer.

  In summary, hypomethylated molecules were shorter in plasma than hypermethylated molecules. The same trend was observed for both fetal and maternal DNA molecules. Given that it is known to affect the packing of DNA methylated nucleosomes, our data indicate that hypomethylated DNA molecules are probably less densely packed with histones, which results in enzymatic degradation. It suggests that they are more susceptible. On the other hand, the data shown in Figures 18A and 18B indicate that the fetal and maternal DNA size distributions are not completely separate from each other, even though fetal DNA is much less methylated than maternal reads. Was also shown. It can be seen in FIG. 19B that the fetal-specific and maternal-specific reads have different methylation levels, even though they are of the same size classification. This observation suggests that the hypomethylated state of fetal DNA is not the only factor explaining its relative shortness to maternal DNA.

VII. Imprinted state of the locus Fetus-derived DNA molecules that share the same genotype as maternal but different epigenetic signatures in maternal plasma are detectable (LLM Poon et al. 2002 Clin Chem; 48: 35- 41). To show that the sequencing approach is sensitive in capturing fetal DNA molecules in maternal plasma, the same strategy was applied to the detection of imprinted fetal alleles in maternal plasma samples. Two genomic imprinted regions have been identified: H19 (chromosome 11: 1,977,419-1977,821, NCBI Build36 / hg18) and MEST (chromosome 7: 129,917,976-129,920). , 347, NCBI Build 36 / hg18). Both of them contain valuable SNPs for distinguishing maternal and fetal sequences. Regarding H19, which is a gene expressed in the mother, for the SNP rs2071094 (chromosome 11: 1,977,740) in the region, the mother is homozygous (A / A) and the fetus is heterozygous (A / C). One of the A maternal alleles was fully methylated and the other was unmethylated. However, in placenta, the allele A was unmethylated, while the C allele inherited from the father was fully methylated. Two methylated leads with a C genotype corresponding to the imprinted paternal allele from placenta were detected in maternal plasma.

  MEST, also known as PEG1, is a paternally expressed gene. Both mother and fetus were heterozygous (A / G) for SNP rs2301335 (Chromosome 7: 129,920,062) within the imprinted locus. In maternal blood, the G allele was methylated while the A allele was unmethylated. The methylation pattern in the placenta was reversed, the maternal A allele was methylated and the paternal G allele was unmethylated. The three unmethylated G alleles from the father were detectable in maternal plasma. In contrast, VAV1, a non-imprinted locus on chromosome 19 (chromosome 19: 6,723, 621-6, 724, 121), shows any allelic methylation pattern in tissue and plasma DNA samples. There wasn't.

  Therefore, the methylation status can be used to determine which DNA fragment is of fetal origin. For example, detection of the A allele in maternal plasma alone cannot be used as a fetal marker when the mother is GA heterozygous. However, when identifying the methylation status of A molecules in plasma, methylated A molecules are fetal-specific, while unmethylated A molecules are maternal-specific and vice versa. .

  Next, we focused on the loci reported to show genomic imprinting in placental tissue. Based on the list of loci reported by Woodfine et al. (2011 Epigenetics Chromatin; 4: 1), loci that contained SNPs within the imprinted control region were further selected. Four loci met the criteria, these were H19, KCNQ10T1, MEST and NESP.

  With respect to the leads of the H19 and KCNQ10T1 maternal blood cell samples, these maternal reads were homozygous for the SNPs with approximately equal proportions of methylated and unmethylated reads. CVS and late pregnancy placental tissue specimens indicate that the fetus is heterozygous for both loci and each allele is exclusively methylated or unmethylated, ie, monoallelic methylation Became clear. In the maternal plasma sample, fetal DNA molecules inherited from the father were detected at both loci. In H19, the paternally inherited molecule was methylated, represented by a sequenced read that contained a fetal-specific allele. In KCNQ10T1, the paternally inherited molecule was represented by a sequenced read that contained a fetal-specific allele and was unmethylated.

  On the other hand, the mother was heterozygous for both MEST and NESP. For MEST, both mother and fetus were GA heterozygotes for SNPs. However, the methylation status of CpGs adjacent to SNPs was opposite in the mother and fetus, as evidenced by data on maternal blood cells and placental tissue Watson chains. The A allele was unmethylated in maternal DNA, but methylated in fetal DNA. For MEST, the maternal allele was methylated. Thus, the fetus has the A allele (methylated in CVS) inherited from its mother, and the mother has the A allele (unmethylated in maternal blood cells) inherited from her father. You can identify what you were doing. Interestingly, in maternal plasma samples, all four molecular groups were easily distinguishable (eg, each of the mother's two alleles and each of the fetus's two alleles). Therefore, fetal DNA molecules inherited from the mother could be easily distinguished from background maternal DNA molecules by combining the genotype information with the methylation status at the imprinted locus (LLM Poon et al. 2002 Clin. Chem; 48: 35-41).

  This approach can be used to detect uniparental disomy. For example, if the father of this fetus is known to be homozygous for the G allele, the failure to detect the unmethylated G allele in maternal plasma indicates a lack of paternal allele contribution. ing. Furthermore, under such circumstances, if both the methylated G allele and the methylated A allele are detected in the plasma of this pregnant woman, the fetus will develop maternal heterodisomy. It is suggested that they have, that is, they do not inherit from their father but inherit two different alleles from their mother. Alternatively, both the methylated A allele (fetal allele inherited from the mother) and the unmethylated A allele (maternal allele inherited from the maternal grandfather) are both unmethylated G alleles. The fetus has maternal-derived isodisomy when it is detected in maternal plasma without the gene (the paternal allele that should have been inherited by the fetus), ie, two identical alleles from the mother He inherited it, and his father suggests he has not inherited anything.

  For NESP, the mother was GA heterozygous at the SNP, while the fetus was homozygous for the G allele. In NESP, the paternal allele was methylated. In maternal plasma samples, the methylated, paternally inherited fetal G alleles were easily distinguishable from the unmethylated background maternal G alleles.

VIII. Cancer / Donors In some embodiments, cancer detection, screening, monitoring (eg, relapse, remission, or response to therapy (eg, present or absent)) using methylation analysis of circulating plasma / serum DNA. ,), Staging, classification (eg, to aid in selecting the most appropriate treatment) and prognosis.

  Cancer DNA is known to exhibit abnormal DNA methylation (JG Herman et al. 2003 N Engl J Med; 349: 2042-2054). For example, compared to non-cancer cells, the CG island promoter of a gene (eg, tumor suppressor gene) is hypermethylated, but the CpG site of the gene itself is hypomethylated. If the methylation profile of cancer cells can be reflected in the methylation profile of tumor-derived plasma DNA molecules using the methods described herein, the overall methylation profile in plasma is cancer-bearing. It is expected that there will be differences between individuals with cancer when compared to healthy individuals or when the cancer has been cured. The type of difference in methylation properties can be in terms of quantitative differences in the methylation density of the genome and / or of the compartments of the genome. For example, due to the universal hypomethylation property of DNA derived from cancer tissues (Gama-Sosa MA et al. 1983 Nucleic Acids Res; 11: 6883-6894), there is a decrease in methylation density in plasma methylome or genomic compartments. , Will be observed in the plasma of cancer patients.

  Quantitative changes in methylation profile should also be reflected among plasma methylome data. For example, a plasma DNA molecule derived from a gene that is hypermethylated only in cancer cells is highly expressed in the plasma of cancer patients when compared to plasma DNA molecules derived from the same gene but in healthy control samples. Will show methylation. Because aberrant methylation occurs in most cancers, the methods described herein can be applied to any form of malignancy with aberrant methylation, including but not limited to lung, breast, colorectal. , Prostate, nasopharynx, stomach, testis, skin, nervous system, bone, ovary, liver, blood system tissue, pancreas, uterus, kidney, bladder, lymphoid tissue and the like. Malignant tumors may be malignant tumors of various histological subtypes, such as, for example, carcinoma, adenocarcinoma, sarcoma, fibroadenocarcinoma, neuroendocrine, and undifferentiated carcinoma.

  On the other hand, it is expected that tumor-derived DNA molecules can be distinguished from background non-tumor-derived DNA molecules, because the overall short size characteristics of tumor-derived DNA have an additional effect on the size of DNA molecules. This is because it is emphasized in DNA molecules derived from loci having abnormal tumor-associated aberrant hypomethylation. Tumor-derived plasma DNA molecules also have multiple characteristics associated with tumor DNA, such as, but not limited to, single nucleotide mutations, increased and decreased copy numbers, translocations, inversions, abnormally high or low methylation. Characterization and size characteristics can be used to distinguish from background non-tumor derived plasma DNA molecules. Since all of these changes can occur independently, the combination of these features may offer additional advantages for sensitive and specific detection of cancer DNA in plasma.

A. Size and Cancer The size of tumor-derived DNA molecules in plasma is also similar to the size of mononucleosome units and is shorter than the background non-tumor-derived DNA molecules that are simultaneously present in plasma of cancer patients. Size parameters have been shown to be associated with cancer, as described in US Patent Application Publication No. 13 / 789,553, incorporated by reference for all purposes.

  Tumor-derived DNA molecules are expected to show the same trend, as both fetal and maternal DNA in plasma showed an association between molecular size and methylation status. For example, a hypomethylated molecule will be shorter than a hypermethylated molecule in the plasma of cancer patients or in a subject undergoing a cancer test.

B. Methylation Densities of Different Tissues in Cancer Patients In this example, plasma and tissue samples of hepatocellular carcinoma (HCC) patients were analyzed. Blood samples were taken from HCC patients before and one week after surgical resection of the tumor. Plasma and buffy coat were collected after centrifugation of blood samples. Excised tumor and adjacent non-neoplastic liver tissue were collected. DNA samples extracted from plasma and tissue samples were analyzed using massively parallel sequencing with and without prior bisulfite treatment. Plasma DNA obtained from 4 healthy individuals without cancer was also analyzed as a control. Bisulfite treatment of DNA samples converts unmethylated cytosine residues to uracil. In the downstream polymerase chain reaction and sequencing, these uracil residues behave as thymidines. On the other hand, bisulfite treatment does not convert methylated cytosine residues to uracil. After massively parallel sequencing, sequencing reads were analyzed using Methy-Pipe (P Jiang, et al. Methy-Pipe: An integrated bioinformatics data analysis pipeline for whole genome methylome analysis, paper presented at the IEEE International. Conference on Bioinformatics and Biomedicine Workshops, Hong Kong, 18 to 21 December 2010), the methylation status of cytosine residues at all CG dinucleotide positions (ie CpG sites) was determined.

  FIG. 21A is a table 2100 showing pre-operative plasma and tissue sample methylation densities of HCC patients. The CpG methylation density of a region of interest (eg, CpG site, promoter, or repeat region) refers to the ratio of reads showing CpG methylation to the total number of reads covering genomic CpG dinucleotides. The methylation densities of buffy coat and non-neoplastic liver tissue were similar. The overall methylation density of tumor tissue based on data obtained from all autosomes was 25% lower than that of buffy coat and non-neoplastic liver tissue. Hypomethylation was consistent on all individual chromosomes. Plasma methylation density was between that of non-malignant and cancerous tissues. This observation is consistent with the fact that both cancerous and non-cancerous tissue contribute to circulating DNA in cancer patients. It has also been shown that the hematopoietic system is the major source of circulating DNA in individuals who do not have an active malignant state (YYN Lui, et al. 2002 Clin Chem; 48: 421-7). Therefore, plasma samples obtained from 4 healthy control groups were also analyzed. The number of sequence reads and sequencing depth achieved for each sample is shown in Table 2150 of Figure 21B.

  FIG. 22 is a table 220 showing that autosomal methylation densities ranged from 71.2% to 72.5% in plasma samples from the healthy control group. These data showed predictive levels of DNA methylation in plasma samples obtained from individuals who had no source of tumor DNA. In cancer patients, tumor tissues also release DNA into the bloodstream (KCA Chan et al. 2013 Clin Chem; 59: 211-224); RJ Leary et al. 2012 Sci Transl Med; 4: 162ra154). Due to the hypomethylated nature of HCC tumors, the presence of both tumor-derived and non-tumor-derived DNA in the patient's pre-operative plasma resulted in reduced methylation density when compared to plasma levels in healthy controls. Bring In fact, the methylation densities of the pre-operative plasma samples were between those of the tumor tissue and the plasma of healthy controls. The reason is that the plasma DNA methylation level of cancer patients is affected by the degree of abnormal methylation of tumor tissue (in this case, hypomethylation) and the concentration fraction of tumor-derived DNA in the circulating blood. Is. The lower methylation density of tumor tissue and the higher concentration fraction of tumor-derived DNA in the circulating blood results in lower methylation density of plasma DNA in cancer patients. Most tumors have been reported to exhibit global hypomethylation (JG Herman et al. 2003 N Engl J Med; 349: 2042-2054; MA Gama-Sosa et al. 1983 Nucleic Acids Res; 11: 6883-6894). Therefore, the current findings found in HCC samples should be applicable to other types of tumors.

  In one embodiment, the methylation density of plasma DNA can be used to determine the concentration fraction of tumor-derived DNA in plasma / serum samples when the methylation level of tumor tissue is known. Tumor tissue methylation levels (eg, methylation density) can be obtained if a tumor sample is available or a tumor biopsy is available. In another embodiment, information about the methylation level of tumor tissue can be obtained from a survey of methylation levels in groups of tumors of similar type, which information (eg, mean or median level) It is applied to patients analyzed using the technology described in. The level of tumor tissue methylation can be determined by analysis of the patient's tumor tissue, or inferred from analysis of tumor tissue of other patients with the same or similar cancer types. Tumor tissue methylation may be performed by a series of methylation recognition platforms such as, but not limited to, massively parallel sequencing, single molecule sequencing, microarrays (eg, oligonucleotide arrays), or mass spectrometry (eg, Epiperper analysis). , Sequenom, Inc.). In some embodiments, such an analysis may be preceded by a methylation status sensitive procedure of the DNA molecule, including but not limited to cytosine immunoprecipitation and methylation recognition restriction enzyme digestion. If the methylation level of the tumor is known, the concentration fraction of tumor DNA in the plasma of cancer patients can be calculated after plasma methylome analysis.

  The relationship between plasma methylation level (P), tumor DNA concentration fraction (f), and tumor tissue methylation level (TUM) may be described as P = BKG × (1-f) + TUM × f. Yes, where BKG is the background DNA methylation level in plasma from blood cells and other internal organs. For example, the total methylation density of all autosomal chromosomes was shown to be 42.9% in tumor biopsy tissues obtained from this HCC patient (ie TUM value in this case). The average methylation density (ie BKG value in this case) of plasma samples obtained from 4 healthy control groups was 71.6%. The plasma methylation density of preoperative plasma was 59.7%. Using these values, f is estimated to be 41.5%.

  In another embodiment, tumor tissue methylation levels can be estimated non-invasively based on plasma methylome data when the concentration fraction of tumor-derived DNA in the plasma sample is known. The concentration fraction of tumor-derived DNA in plasma samples can be determined by other genetic analyzes such as those described above (US Patent Application Publication No. 13 / 308,473; KCA Chan et al. 2013 Clin Chem; 59: 211-24). , Allele deletion (GAAL) genome-wide analysis and single nucleotide mutation analysis. The calculations are based on the above relationships, which in this embodiment are the same, except that the value of f is known and the value of TUM is unknown. Estimates can be made for the entire genome, or for portions of the genome, similar to the data observed in the context of determining placental tissue methylation levels from maternal plasma data.

  In another embodiment, in-bin variation or characteristics of methylation density can be utilized to distinguish between subjects with and without cancer. The resolution of methylation analysis can be further increased by dividing the genome into bins of specific size (eg, 1 Mb). In such embodiments, methylation densities of each 1 Mb bin are collected before and after harvested samples, eg, buffy coat, excised HCC tissue, non-neoplastic liver tissue adjacent to the tumor, and tumor resection. Calculated for plasma. In another embodiment, the bin size may not be constant. In some implementations, the bins themselves may differ in size, but the number of CpG sites is constant within each bin.

  23A and 23B show the methylation densities of buffy coat, tumor tissue, non-neoplastic liver tissue, pre- and post-operative plasma of HCC patients. FIG. 23A is a plot 2300 of results on chromosome 1. FIG. 23B is a plot 2350 of the results on chromosome 2.

  In the majority of the 1 Mb windows, the methylation densities of buffy coat and non-neoplastic liver tissue adjacent to the tumor were similar, while tumor tissue had a lower methylation density. Preoperative plasma methylation density is between that of tumor and non-malignant tissue. Methylation densities of matched genomic regions within tumor tissue can be estimated using preoperative plasma methylation data and tumor DNA concentration fractions. The method is the same as above using all autosomal methylation density values. The described tumor methylation estimation can also be performed using this higher resolution plasma DNA methylation data. Other bin sizes such as 300 kb, 500 kb, 2 Mb, 3 Mb, 5 Mb or more than 5 Mb can also be used. In one embodiment, the bin size may not be constant. In some implementations, the bins themselves may differ in size, but the number of CpG sites is constant within each bin.

C. Comparison of Plasma Methylation Density Between Cancer Patients and Healthy Subjects As shown in 2100, the methylation density of preoperative plasma DNA was lower than that of non-malignant tissues in cancer patients. This may be due to the presence of tumor tissue-derived DNA that was in a hypomethylated state. This lower plasma DNA methylation density can potentially be used as a biomarker for cancer detection and monitoring. In cancer monitoring, when the cancer is in progress, an increase in the amount of cancer-derived DNA in plasma occurs over time. In this example, an increase in the amount of circulating cancer-derived DNA in plasma leads to a further decrease in plasma DNA methylation density at genome-wide levels.

  Conversely, if the cancer is responding to treatment, the amount of cancer-derived DNA in plasma will decrease over time. In this example, a decrease in the amount of cancer-derived DNA in plasma leads to an increase in plasma DNA methylation density. For example, when lung cancer patients with epidermal growth factor receptor mutations are treated by targeted therapy (eg, tyrosine kinase inhibition), increased plasma DNA methylation density is indicative of a response. Subsequent development of tumor clones resistant to tyrosine kinase inhibition is associated with a decrease in plasma DNA methylation density indicative of relapse.

  Plasma methylation density measurements can be performed continuously and the rate of change of such measurements can be calculated and used to predict or correlate clinical progression or amelioration or prognosis. At selected genomic loci that are hypermethylated in cancer tissues but hypomethylated in normal tissues (eg, the promoter region of some tumor suppressor genes) for the progression and treatment of cancer. The relevance between the preferred responses of the is opposite to the above pattern.

  To demonstrate the feasibility of this approach, DNA methylation densities of plasma samples taken from cancer patients before and after surgical resection of tumors were compared with plasma DNA obtained from four healthy control patients. did.

  Table 2200 shows a combination of DNA methylation densities and autosomal values for each autosomal chromosome in all pre- and post-surgical plasma samples of cancer patients and in 4 healthy healthy control patients. On all chromosomes, the methylation densities of pre-operative plasma DNA samples were lower than those of post-operative samples and plasma samples obtained from 4 healthy subjects. Differences in plasma DNA methylation densities between pre- and post-operative samples provided supporting evidence that the lower methylation densities in pre-operative plasma samples were due to the presence of HCC tumor-derived DNA. To be

  Restoration of DNA methylation density in post-operative plasma samples to levels similar to those in healthy control plasma samples suggests that the majority of tumor-derived DNA was lost by surgical excision of the source (ie tumor) Was done. These data indicate that methylation density in pre-operative plasma is lower than that in healthy controls, as determined using data available from large genomic regions (eg, whole autosomes or individual chromosomes). Level, which suggests that it is possible to identify (ie, diagnose or screen) a laboratory case with cancer.

  Methylation levels even lower than post-operation plasma methylation levels can also be used to monitor tumor burden, thereby prognosing and monitoring cancer progression in patients Was also shown by preoperative plasma data. Reference values can be determined from the plasma of healthy controls or those at risk of cancer but not currently having cancer. Those at risk for HCC include those with a chronic hepatitis B or C infection, those with hemochromatosis, and those with cirrhosis.

  Plasma methylation density values above a predetermined cutoff (eg, lower) based on baseline values can be used to assess whether plasma of non-pregnant women has tumor DNA. In order to detect the presence of circulating tumor DNA in a hypomethylated state, the cutoff is below the 5 or 1 percentile of the value of the control population, or the number of standard deviations below the control mean methylated density value, eg 2 Alternatively, it can be defined based on a standard deviation (SD) of 3 or based on the determination of median multiple (MoM). For hypermethylated tumor DNA, the cutoff is higher than the 95th or 99th percentile of the control population value, or at a number of standard deviations above the control mean methylated density value, eg, SD of 2 or 3. Or based on the determination of median multiple (MoM). In one embodiment, the control population is age-matched with the test subject. Age matching need not be exact and may occur in the age range (eg, 30-40 years for a 35-year-old test subject).

  Next, the methylation densities of 1 Mb bins were compared between plasma samples of cancer patients and 4 control patients. For illustration, the results for chromosome 1 are shown.

  FIG. 24A is a plot 2400 showing the methylation density of pre-operative plasma from HCC patients. FIG. 24B is a plot 2450 showing the methylation density of post-operative plasma from HCC patients. Blue dots represent the results for control patients and red dots represent the results for plasma samples from HCC patients.

  As shown in Figure 24A, the methylation density of pre-operative plasma from HCC patients was lower in most bins than that of control patients. Similar patterns were observed on other chromosomes. As shown in Figure 24B, the methylation density of post-operative plasma from HCC patients was similar to that of control patients in most bins. Similar patterns were observed on other chromosomes.

  The results of the test subjects were compared to the values of the reference group to assess whether the test subjects had cancer. In one embodiment, the reference group consists of some healthy subjects. In another embodiment, the reference group may consist of subjects having a non-malignant condition (eg, chronic hepatitis B infection or cirrhosis). The difference in methylation density between the test subject and the reference group can then be quantified.

  In one embodiment, the reference range can be obtained from the control values. Deviations in the test subject's results from the upper or lower limits of the reference group can then be used to determine if the subject has a tumor. This amount is affected by the concentration fraction of tumor-derived DNA in plasma and the difference in methylation levels between malignant and non-malignant tissues. The higher the concentration fraction of tumor-derived DNA in plasma, the greater the methylation density difference between the test plasma sample and the control. The greater degree of difference in methylation levels in malignant and non-malignant tissues is also associated with greater differences in methylation density between test plasma samples and controls. In yet another embodiment, different reference groups were chosen for subjects of different age ranges.

  In another embodiment, the mean methylation density and SD of 4 control patients were calculated for each 1 Mb bin. Then, for the corresponding bins, the difference between the methylation density of HCC patients and the mean value of control patients was calculated. In one embodiment, this difference was then divided by the SD of the corresponding bins to determine the Z value. In other words, the Z value represents the difference in methylation density between the test and control plasma samples, expressed as the number of SDs from the mean of control patients. A Z value greater than 3 in a bin indicates that the plasma DNA of HCC patients is more highly methylated in that bin with an SD greater than 3 while that in a bin is less than -3 in a bin. Z values indicate that the plasma DNA of HCC patients is hypomethylated in their bins by SD greater than 3 than control patients.

  25A and 25B are plasma DNA of plasma samples of pre-operative (Plot 2500) and post-operative (Plot 2550) HCC patients using plasma methylome data from four healthy control patients as a reference to chromosome 1. The Z value of the methylation density is shown. Each dot represents the result of one 1 Mb bin. Black dots represent bins with Z values of -3 to 3. Red dots represent bins with Z values less than -3.

  FIG. 26A is a table 2600 showing Z value data for pre- and post-operative plasma. The majority of the bins on chromosome 1 (80.9%) in the pre-operative plasma samples had Z-values less than -3 because pre-operative plasma DNA in HCC patients was pre-operative in control patients. This indicates that the methylation state was significantly lower than that of plasma DNA. On the contrary, the number of red dots was substantially reduced in post-operative plasma samples (8.3% of bins on chromosome 1), which was due to the majority of the tumor DNA as a source of circulating tumor DNA. It has been removed from circulation due to surgical resection.

  FIG. 26B is a Circos showing Z-values of plasma DNA methylation densities of pre-operative and post-operative plasma samples of HCC patients, using 4 healthy control patients as a reference for 1 Mb bins analyzed from all autosomes. This is plot 2620. The outermost ring indicates the human autosomal symbol. The middle ring shows data for pre-operative plasma samples. The innermost ring shows data for post-operative plasma samples. Each dot represents the result of one 1 Mb bin. Black dots represent bins with Z values of -3 to 3. Red dots represent bins with Z values less than -3. Green dots represent bins with Z values greater than 3.

  FIG. 26C is a table 2640 showing the distribution of whole genome 1 Mb bin Z values in both pre- and post-operative plasma samples of HCC patients. The results show that pre-operative plasma DNA of HCC patients was less methylated than control pre-operative plasma DNA in most of the regions in the whole genome (85.2% of 1 Mb bin). There is. In contrast, the majority of the areas in post-operative plasma samples (93.5% of 1 Mb bins) did not show significant hypermethylation or hypomethylation compared to controls. These data indicate that most of the naturally predominantly hypomethylated tumor DNA of this HCC was not present in post-operative plasma samples.

  In one embodiment, the number, percentage or ratio of bins with Z values less than -3 can be used to indicate whether cancer is present. For example, as shown in Table 2640, 2330 (85.2%) of the 2734 bins analyzed showed a Z value of less than -3 in preoperative plasma, but 171 of the 2734 bins analyzed. Only (6.3%) showed a Z value of less than -3 in post-operative plasma. The data show that the amount of tumor DNA in pre-operative plasma was significantly higher than the amount of tumor DNA in post-operative plasma.

  The cutoff value for the number of bottles can be determined using statistical methods. For example, approximately 0.15% of bins are predicted to have Z-values of less than -3 based on the normal distribution. Therefore, the cutoff number of bins may be 0.15% of the total number of bins being analyzed. In other words, if a plasma sample obtained from a non-pregnant individual exhibits more than 0.15% bins with a Z value of less than -3, the source of hypomethylated DNA in plasma, i.e. the cancer is Exists. For example, 0.15% of the 2734 1 Mb bins analyzed in this example are approximately 4 bins. Using this value as a cut-off, both pre- and post-surgical plasma samples contained DNA from hypomethylated tumors, but the amount was higher in pre-surgical plasma samples than in post-surgical plasma samples. It was quite a lot. None of the bins showed significant hypermethylation or hypomethylation in 4 healthy control patients. Other cutoff values (eg, 1.1%) can also be used and can vary depending on the requirements of the assay used. As another example, the cutoff rate may vary depending on the statistical distribution and the desired sensitivity and acceptable specificity.

  In another embodiment, the cutoff value may be determined by receiver operating characteristic (ROC) curve analysis by analyzing some cancer patients and individuals without cancer. To further confirm the specificity of this approach, plasma samples obtained from patients seeking non-malignant status (C06) consultation were analyzed. 1.1% of the bottles had a Z value of less than -3. In one embodiment, different thresholds may be used to classify different levels of disease states. A lower threshold rate can be used to distinguish a healthy state from a benign condition, and a higher threshold rate can be used to distinguish a benign condition from a malignant tumor.

  The diagnostic ability of plasma hypomethylation analysis using massively parallel sequencing uses polymerase chain reaction (PCR) -based amplification of specific classes of repeat regions (eg, long interspersed repeat sequence-1 (LINE-1)). It appears to be superior to that obtained by P Tangkijvanich et al. 2007 Clin Chim Acta; 379: 127-133. One possible explanation for this finding is that hypomethylation is widespread in the tumor genome, but with some heterogeneity between genomic regions.

  Indeed, it was observed that the mean plasma methylation densities of the reference subjects differ across the genome (Figure 56). Each red dot in FIG. 56 represents the average methylation density of one 1 Mb bin among 32 healthy subjects. The plot shows all 1 Mb bins analyzed across the genome. The number in each box represents the number of chromosomes. It was observed that the average methylation density varied from bottle to bottle.

  A simple PCR-based assay would not be able to take such region-wise heterogeneity into account in its diagnostic algorithm. Such heterogeneity broadens the range of methylation densities observed in healthy individuals. Larger reductions in methylation density are required for samples considered to exhibit hypomethylation. This results in a reduction in test sensitivity.

  In contrast, the massively parallel sequencing-based approach divides the genome into 1 Mb bins (or bins of other sizes) and measures the methylation density of such bins individually. This approach reduces the effect of variations in baseline methylation density over different genomic regions, as such regions are compared between test samples and controls. In fact, within the same bin, the inter-individual variability between the 32 healthy control groups was relatively small. 95% of the bins had a coefficient of variation (CV) of 1.8% or less among the 32 healthy control groups. It should be noted that the comparison can be performed across multiple genomic regions in order to further enhance the sensitivity of detection of hypomethylation associated with cancer. Sensitivity is enhanced by testing multiple genomic regions, as it protects against the effects of biological variation when a cancer sample does not accidentally show hypomethylation in a specific region when only one region is tested It will be done.

  An approach to compare the methylation densities of corresponding genomic regions between control and test samples (eg, examine each genomic region separately and then combine such results) and make this comparison for multiple genomic regions is , With a higher signal-to-noise ratio in the detection of cancer-related hypomethylation. This massively parallel sequencing approach is shown as an example. Other methodologies that can determine the methylation densities of multiple genomic regions and allow comparison of the methylation densities of corresponding regions between control and test samples are expected to achieve similar effects. For example, hybridization or molecular inversion probes that can target plasma DNA molecules from a particular genomic region and determine the methylation level of that region are designed to achieve the desired effect. can do.

  In yet another embodiment, the sum of the Z values of all bins can be used to determine if cancer is present or to monitor continuous changes in the level of plasma DNA methylation. Due to the overall hypomethylated nature of tumor DNA, the sum of Z values will be lower in plasma collected from individuals with cancer than in healthy controls. The sum of the Z values of the plasma samples of HCC patients before and after surgery was -4983.8 and -132.13, respectively.

  In other embodiments, other methods can be used to determine the methylation level of plasma DNA. For example, the ratio of methylated cytosine residues to the total amount of cytosine residues can be determined using mass spectrometry (ML Chen et al. 2013 Clin Chem; 59: 824-832) or massively parallel sequencing. . However, since the majority of cytosine residues are not present in the CpG dinucleotide sequence, the percentage of methylated cytosines in all cytosine residues is relatively small when compared to the estimated methylation level in the CpG dinucleotide sequence. It will be. The methylation levels of tissue and plasma samples obtained from HCC patients and four plasma samples obtained from healthy controls were determined. Genome-wide massively parallel sequencing data was used to measure methylation levels in CpG sequences, any cytosine, CHG and CHH sequences. H refers to adenine, thymine or cytosine residues.

  FIG. 26D is a table 2660 showing methylation levels of tumor tissue and pre-operative plasma samples that overlap some of the control plasma samples with the CHH and CHG sequences. Methylation levels of tumor tissue and pre-operative plasma samples were consistently lower in both CpG and nonspecific cytosine compared to buffy coat, non-neoplastic liver tissue, post-operative plasma samples and healthy control plasma samples. However, the data based on methylated CpG, ie methylation density, showed a wider dynamic range than the data based on methylated cytosine.

  In other embodiments, the methylation status of plasma DNA can be determined by methods using antibodies to methylated cytosine, such as methylated DNA immunoprecipitation (MeDIP). However, the accuracy of these methods is expected to be less than that of sequencing-based methods due to variability in antibody binding. In yet another embodiment, the level of 5-hydroxymethylcytosine in plasma DNA can be determined. In this regard, reduced levels of 5-hydroxymethylcytosine have been found to be an epigenetic feature of certain cancers (eg, melanoma) (CG Lian, et al. 2012 Cell; 150: 1135-1146).

  In addition to HCC, we also investigated whether this approach is adaptable to other cancer types. 2 patients with lung adenocarcinoma (CL1 and CL2), 2 patients with nasopharyngeal cancer (NPC1 and NPC2), 2 patients with colorectal cancer (CRC1 and CRC2), metastatic neuroendocrine Plasma samples obtained from one patient with tumor (NE1) and one patient with metastatic leiomyosarcoma (SMS1) were analyzed. Plasma DNA of these subjects was converted with bisulfite and 50 bp at one end was sequenced using the Illumina HiSeq 2000 platform. The 4 healthy control patients described above were used as a reference group for the analysis of these 8 patients. A 50 bp sequence read at one end was used. The entire genome was divided into 1 Mb bins. The data obtained from the reference group was used to calculate the mean and SD of the methylation densities for each bin. Next, the results of 8 cancer patients were expressed as Z values showing the numerical value of SD from the average value of the reference group. A positive value indicates that the methylation density of the test case is lower than the mean value of the reference group, and a negative value indicates that the methylation density of the test case is higher than the mean value of the reference group. . The number of sequence reads and the sequencing depth achieved for each sample is shown in Table 2780 of Figure 27I.

  27A-H are Circos plots of methylation densities of eight cancer patients, according to embodiments of the invention. Each dot represents a 1 Mb bin result. Black dots represent bins with Z values of -3 to 3. Red dots represent bins with Z values less than -3. Green dots represent bins with Z values greater than 3. The spacing between two continuous lines represents a Z value difference of 20.

  Significant hypomethylation was observed in multiple regions across the genome of patients with most cancer types including lung, nasopharyngeal, colorectal and metastatic neuroendocrine tumors. Interestingly, in addition to hypomethylation, significant hypermethylation was observed in multiple regions throughout the genome in cases with metastatic leiomyosarcoma. The embryonic origin of leiomyosarcoma is mesoderm, while the embryonic origin of the other cancer types in the remaining 7 patients is ectoderm. Therefore, the DNA methylation pattern of sarcoma can be different from that of carcinoma.

  As can be seen from this example, the methylation pattern of plasma DNA can be useful in distinguishing different types of cancer, in this case distinguishing between carcinoma and sarcoma. These data also suggest that the approach can be used to detect aberrant hypermethylation associated with malignancy. In all eight of these cases, only plasma samples were available and no tumor tissue was analyzed. This indicates that tumor-derived DNA can be easily detected in plasma using the described method, even without prior methylation properties or levels of methylation in the tumor tissue.

  FIG. 27J is a table 2790 showing the Z value distribution of 1 Mb bins throughout the genome in plasma of patients with different malignancies. The percentage of bins with Z values less than -3, -3 to 3 and greater than 3 are shown for each example. In all cases, greater than 5% of the bins had Z values less than -3. Therefore, if we use the cutoff that 5% of the bins are significantly hypomethylated to classify a sample as cancer positive, then all of these examples would be classified as cancer positive. . The results indicate that hypomethylation may be a common phenomenon in various cancer types and plasma methylome analysis may be useful in detecting various cancer types.

D. Method FIG. 28 is a flow chart of a method 2800 of analyzing a biological sample of an organism to determine a cancer level classification in accordance with an embodiment of the present invention. A biological sample contains DNA from normal cells and may potentially contain DNA from cells associated with cancer. At least a portion of the DNA can be cell-free DNA in the biological sample.

  At block 2810, a plurality of DNA molecules from the biological sample are analyzed. Analyzing a DNA molecule can include determining the location of the DNA molecule within the organism's genome and determining whether the DNA molecule is methylated at one or more sites. Since the analysis can be performed by obtaining a sequence read from methylation recognition sequencing, the analysis can only be performed on the data previously obtained from the DNA. In other embodiments, the analysis may include active steps to obtain actual sequencing or other data.

  At block 2820, for each of the plurality of sites, the number of each DNA molecule that is methylated at the site is determined. In one embodiment, the site is a CpG site, and may only be one particular CpG site selected using one or more of the criteria described herein. The number of DNA molecules that are methylated is determined by determining the number that is unmethylated if normalization was performed using the total number of DNA molecules analyzed at a particular site (eg, total number of sequence reads). be equivalent to. For example, an increase in CpG methylation density in a region is equivalent to a decrease in unmethylated CpG density in the same region.

  At block 2830, a first methylation level is calculated based on the respective numbers of DNA molecules that are methylated at the plurality of sites. The first methylation level can correspond to a methylation density determined based on the number of DNA molecules corresponding to multiple sites. A site can correspond to multiple loci or only one locus.

  At block 2840, the first methylation level is compared to the first cutoff value. The first cutoff value may be the reference methylation level or it may be associated with the reference methylation level (eg, a certain distance from normal levels). Reference methylation levels can be determined from samples of individuals who do not have cancer, or from genetic loci of organisms known not to be associated with cancer of the organism. The first cutoff value may be established from a reference methylation level obtained from a previous biological sample of the organism, obtained prior to examination of the biological sample.

  In one embodiment, the first cutoff value is a particular distance (eg, a particular value of standard deviation) from a reference methylation level established from a biological sample obtained from a healthy organism. The comparison can be performed by determining the difference between the first methylation level and the reference methylation level and then comparing the difference to a threshold corresponding to the first cutoff value (eg, methylation level is (To determine if it is statistically different from the reference methylation level).

  At block 2850, a classification of cancer levels is determined based on the comparison. Examples of levels of cancer include whether the subject has cancer or a precancerous condition, or whether the subject has an increased likelihood of developing cancer. In one embodiment, the first cutoff value can be determined from a sample previously obtained from the subject (eg, the reference methylation level can be determined from the previous sample).

  In some embodiments, the first methylation level can correspond to the number of regions where the methylation level is above the threshold. For example, multiple regions of the organism's genome may be identified. Regions can be identified using the criteria described herein (eg, having a certain length or a certain number of sites). One or more sites (eg, CpG sites) can be identified within each region. Regional methylation levels can be calculated for each region. The first methylation level is for the first region. Each regional methylation level is compared to a respective regional cutoff value, which may be the same or different between regions. The area cutoff value of the first area is the first cutoff value. Since each region cutoff value is a certain amount (eg, 0.5) from the reference methylation level, only regions with significant differences from the reference that can be determined from non-cancer subjects can be counted.

  A first number of regions with regional methylation levels above their respective regional cutoff values can be determined and compared to a threshold to determine a classification. In one implementation, the threshold is a percentage. The comparison of the first number of thresholds involves dividing the first number of regions by the second number of regions (e.g., all regions) and then comparing with the threshold, for example, as part of the normalization process. May be included.

  As described above, the concentration fraction of tumor DNA in the biological sample can be used to calculate the first cutoff value. It can simply be estimated that the concentration fraction is greater than the minimum value, whereas samples with concentration fractions lower than the minimum value can be warned, eg as not suitable for analysis. The minimum value can be determined based on the expected difference in the methylation level of the tumor compared to the reference methylation level. For example, if the difference is 0.5 (eg, used as a cutoff value), then the concentration of a particular tumor needs to be high enough to see this difference.

  Certain techniques from method 1300 can be applied to method 2800. In method 1300, copy number variation can be determined in the tumor (eg, it can be examined whether the chromosome 1 region of the tumor has a copy number change as compared to the chromosome 2 region of the tumor). Therefore, the method 1300 may presume that a tumor is present. In method 2800, the sample can be examined for signs of any tumor present, regardless of any copy number characteristics. Some approaches of the two methods may be similar. However, the cut-off values and methylation parameters (eg, normalized methylation levels) of method 2800 indicate that a mixture of cancer DNA and non-cancerous DNA with some regions that may have copy number changes. In contrast to the difference from the reference methylation level, a statistically significant difference from the reference methylation level of non-cancerous DNA can be detected. Accordingly, the baseline value for Method 2800 is from a cancer-free sample, eg, from a cancer-free organism, or from non-cancerous tissue of the same patient (eg, pre-collected plasma or cells). (Determined from DNA, from contemporaneously obtained samples known not to have cancer).

E. Prediction of Minimum Tumor DNA Concentration Fractions Detected Using Plasma DNA Methylation Analysis One method of measuring the sensitivity of the cancer detection approach using plasma DNA methylation levels is to control plasma DNA methylation levels. It is associated with the minimum tumor-derived DNA concentration fraction required to account for changes in plasma DNA methylation levels when compared to. The test sensitivity also depends on the degree of difference in DNA methylation between tumor tissue and baseline plasma DNA methylation levels in healthy control or blood cell DNA. Blood cells are the major source of DNA in the plasma of healthy people. The greater the difference, the more easily cancer patients can be distinguished from non-cancerous individuals, reflected in lower detection limits of tumor-derived in plasma and higher clinical sensitivity in detecting cancer patients. Will Furthermore, fluctuations in plasma DNA methylation in healthy subjects or subjects of different ages (G Hannum et al. 2013 Mol Cell; 49: 359-367) also influence the sensitivity of detecting methylation changes associated with the presence of cancer. will do. The smaller the fluctuations in plasma DNA methylation in healthy subjects, the easier it will be to detect the changes caused by the presence of small amounts of cancer-derived DNA.

  FIG. 29A is a plot 2900 showing the distribution of methylation densities in a reference subject, assuming that the distribution follows a normal distribution, this analysis shows one methylation density value (eg, for all autosomal or specific chromosomes). (Methylation density) only, based on each plasma sample. It explains how the specificity of the analysis is affected. In one embodiment, a cutoff of 3 SD below the average DNA methylation density of the reference subject is used to determine if the test sample is significantly less methylated than the sample obtained from the reference subject. If this cutoff is used, approximately 0.15% of non-cancerous subjects will have false positive results classified as having cancer, resulting in a specificity of 99.85%. is expected.

  FIG. 29B is a plot 2950 showing the distribution of methylation densities in reference subjects and cancer patients. The cut-off value is 3 SD below the mean value of the methylation density of the reference. If the mean methylation density of cancer patients is 2 SD below the cutoff value (ie, 5 SD below the mean of the reference subjects), 97.5% of cancer subjects will have a methylation density below the cutoff value. Expected to have. In other words, if one methylation density value is given to each subject, eg the whole methylation density of the whole genome, all autosomes or a particular chromosome is analyzed, the expected sensitivity will be 97.5%. Let's do it. The difference between the average methylation densities of the two populations depends on two factors: the degree of difference in methylation levels between cancerous and non-cancerous tissues, and the concentration of tumor-derived DNA in plasma samples. Affected by fractions. The higher the value of these two parameters, the greater the difference in the methylation density values of these two populations. Furthermore, the lower the SD of the methylation density distribution of the two populations, the less overlap of the methylation density distribution of the two populations.

  Below, this concept is explained using a hypothetical example. It is assumed that the tumor tissue has a methylation density of approximately 0.45 and the plasma DNA of a healthy subject has a methylation density of approximately 0.7. These hypothesized values are from HCC patients whose autosomal total methylation density was 42.9% and the plasma samples obtained from the healthy control group had an autosomal average methylation density of 71.6%. It is similar to the value obtained. Assuming a CV measuring plasma DNA methylation density of the entire genome of 1%, the cut-off value is 0.7 × (100% −3 × 1%) = 0.679. To achieve a sensitivity of 97.5%, the average methylation density of plasma DNA of cancer patients should be approximately 0.679-0.7 x (2 x 1%) = 0.665. . f represents the concentration fraction of tumor-derived DNA in the plasma sample. f can be calculated as (0.7−0.45) × f = 0.7−0.665. From this, f is approximately 14%. From this calculation, the minimum detectable fraction in plasma is 14%, which results in a diagnostic sensitivity of 97.5% when the overall genome-wide methylation density is used as a diagnostic parameter. Estimated to be achieved.

  This analysis was then performed on the data obtained from HCC patients. To account for this, only one methylation density measurement was performed on each sample based on the values estimated from all autosomes. The average methylation density between plasma samples obtained from healthy subjects was 71.6%. The SD of the methylation density of these four samples was 0.631%. Therefore, the cut-off value for plasma methylation density is 71.6% −3 × 0.631% = 69.7% to achieve a Z value of less than −3 and a specificity of 99.85%. There is a need. To achieve a sensitivity of 97.5%, the average plasma methylation density of cancer patients needs to be 2SD below the cutoff (ie, 68.4%). The methylation density of the tumor tissue was 42.9%, so using the formula: P = BKG × (1-f) + TUM × f, f should be at least 11.1%.

  In another embodiment, the methylation densities of different genomic regions can be analyzed separately, eg, as shown in FIG. 25A or FIG. 26B. In other words, multiple methylation level measurements are made on each sample. As shown below, significant hypomethylation can be detected at even lower tumor DNA concentration fractions in plasma, thus enhancing the diagnostic power of plasma DNA methylation analysis in cancer detection. The number of genomic regions showing a significant deviation of methylation density from the reference population can be counted. The number of genomic regions is then compared to a cutoff value to determine if there is an overall significant hypomethylation of plasma DNA over the population of examined genomic regions (eg, 1 Mb bins of the entire genome). Can be determined. The cut-off value is established by analysis of a panel of cancer free reference subjects or is obtained mathematically, for example according to a normal distribution function.

  FIG. 30 is a plot 3000 showing the distribution of methylation densities of plasma DNA in healthy subjects and cancer patients. The methylation density of each 1 Mb bin is compared to the corresponding value in the reference group. The percentage of bins showing significant hypomethylation (3 SD below the mean of the reference group) was determined. A cut-off of 10%, a significant hypomethylation state, was used to determine if tumor-derived DNA was present in plasma samples. 5%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 60%, 70%, 80% or 90% depending on the desired sensitivity and specificity of the test Other cutoff values such as

  For example, to classify a sample as containing tumor-derived DNA, a 1 Mb bin showing 10% significant hypomethylation (Z value less than -3) can be used as a cutoff. A sample is classified as positive in a cancer test if greater than 10% of the bins are significantly more hypomethylated than the reference group. A cutoff of 3 SD below the mean methylation density of the reference group in each 1 Mb bin is used to define the sample as being significantly more hypomethylated. For each 1 Mb bin, if the mean plasma DNA methylation density of the cancer patient is 1.72 SD lower than the mean plasma DNA methylation density of the reference subject, the methylation density value of any particular bin of the cancer patient will be cut There is a 10% probability of giving a positive result, lower than off (ie, Z value less than −3). Next, when examining all 1 Mb bins across the genome, approximately 10% of the bins are expected to show positive results with significantly lower methylation densities (ie, Z values less than -3). . Assuming that the total methylation density of plasma DNA of a healthy subject is about 0.7, and the coefficient of variation (CV) of the measurement of plasma DNA methylation density of each 1 Mb bin is 1%, the The average methylation density would need to be 0.7 x (100% -1.72 x 1%) = 0.68796. This average plasma DNA methylation density is obtained, where f is the concentration fraction of tumor-derived DNA in plasma. Assuming that the methylation density of the tumor tissue is 0.45, f can be calculated using the formula:

In the formula,

Represents the mean methylation density of plasma DNA in the reference individual;

Represents the methylation density of the tumor tissue of a cancer patient;

Represents the average methylation density of plasma DNA of cancer patients.

  Using this equation, (0.7-0.45) xf = 0.7-0.68796. Therefore, the minimum concentration fraction can be detected using this approach and is estimated to be 4.8%. Sensitivity can be further enhanced by reducing the cutoff rate for bins that are significantly more hypomethylated, for example from 10% to 5%.

  As demonstrated in the examples above, the sensitivity of this method is determined by the degree of difference in methylation levels between cancerous and non-cancerous tissues (eg blood cells). In one embodiment, only those chromosomal regions that show large differences in methylation density between plasma DNA and tumor tissue of non-cancerous subjects are selected. In one embodiment, only regions with a difference in methylation density greater than 0.5 are selected. In other embodiments, a difference of 0.4, 0.6, 0.7, 0.8 or 0.9 can be used to select the appropriate region. In yet another embodiment, the physical size of the genomic region is not fixed. Instead, genomic regions are defined based on, for example, a fixed reading depth or a fixed number of CpG sites. Methylation levels in multiples of these genomic regions are evaluated in each sample.

  FIG. 31 is a graph 3100 showing the distribution of differences in methylation density between plasma DNA of healthy subjects and mean values of tumor tissue of HCC patients. Positive values indicate that the methylation density is higher in plasma DNA of healthy subjects, negative values indicate that the methylation density is higher in tumor tissue.

  In one embodiment, the bins that have the greatest difference between the methylation densities of cancerous and non-cancerous tissues (eg, those that have a difference of greater than 0.5) have tumors that are hypomethylated in these bins. It can be selected regardless of state or hypermethylation state. The detection limit of the concentration fraction of tumor-derived DNA in plasma was increased by focusing on these bins due to the greater difference between the distribution of plasma DNA methylation levels between cancer and non-cancer subjects. It can be reduced and tumor-derived DNA in plasma can be given the same concentration fraction. For example, if only bins with a difference of more than 0.5 are used, and a cutoff of 10% of the bins is significantly more hypomethylated is used to determine whether a test individual has cancer. The minimum concentration fraction (f) of tumor-derived DNA detected can be calculated using the following formula:

In the formula,

Represents the mean methylation density of plasma DNA in the reference individual;

Represents the methylation density of tumor tissue in cancer patients;

Represents the mean methylation density of plasma DNA in cancer patients.

  On the other hand, the difference in methylation density between the reference plasma and the tumor tissue is at least 0.5. At that time, 0.5 × f = 0.7−0.68796, and f = 2.4%. Therefore, by focusing on bins with greater differences in methylation densities between cancerous and non-cancerous tissues, the lower limit of tumor-derived DNA fraction was reduced from 4.8% to 2.4%. obtain. Information on which bins show a greater degree of methylation difference between cancerous and non-cancerous tissues (eg blood cells) can be found in tumor tissues of the same organ or of the same tissue type obtained from other individuals. Can be determined.

  In another embodiment, the parameter is derived from the methylation density of plasma DNA in all bins, and differences in methylation density between cancerous and non-cancerous tissues can be considered. Bins with greater differences may be given heavier weight. In one embodiment, the difference in methylation densities between the cancerous and non-cancerous tissues of each bin can be used directly as the weight for a particular bin in calculating the final parameters.

  In yet another embodiment, different cancer types may have different patterns of methylation in tumor tissue. Cancer-specific weight profiles can be derived from the degree of methylation of a particular cancer type.

  In yet another embodiment, the association between methylation density bins can be determined in subjects with and without cancer. In FIG. 8, it can be observed that in a small number of bins, the tumor tissue was more methylated than the reference plasma DNA. Therefore, the bin with the most extreme difference value (eg, greater than 0.5 difference and less than 0 difference) may be selected. The ratio of the methylation densities of these bins can then be used to indicate whether the test individual has cancer. In other embodiments, the difference and quotient of methylation densities of different bins can be used as a parameter to indicate the association between bins.

  The sensitivity of the approach for detecting or assessing tumors using the methylation densities of multiple genomic regions, demonstrated by the data obtained from HCC patients, was further evaluated. First, reads from pre-operative plasma were mixed with reads from healthy control plasma samples to mimic plasma samples containing tumor DNA concentration fractions ranging from 20% to 0.5%. The percentage of 1 Mb bins (out of 2,734 bins in the whole genome) with a methylation density equal to a Z value of less than -3 was then counted. 80.0% of the bins showed significant hypomethylation when the tumor DNA concentration fraction in plasma was 20%. Data corresponding to tumor DNA concentration fractions in plasma of 10%, 5%, 2%, 1% and 0.5% are 67.6%, 49.7%, 18.9%, 3. The bottles showed hypomethylation of 8% and 0.77%. Since the theoretical limit for the number of bins exhibiting Z-values less than -3 in the control sample is 0.15%, our data show that even if the tumor concentration fraction was only 0.5%, the theoretical It indicates that there were more bins (0.77%) above the cutoff limit.

  FIG. 32A is a table 3200 showing the effect of reduced sequencing depth when plasma samples contained 5% or 2% tumor DNA. A high percentage of bins (> 0.15%) that show significant hypomethylation can be detected even when the average sequencing depth was only 0.022 times per haploid genome.

  FIG. 32B is a graph 3250 showing methylation densities of repetitive and non-repetitive regions in plasma of four healthy control patients, buffy coat, normal liver tissue, tumor tissue of HCC patients, pre-operative and post-operative plasma samples. Is. It can be observed that the repetitive regions were more methylated (higher methylation density) than the non-repetitive regions in both cancerous and non-cancerous tissues. However, the difference in methylation between repetitive and non-repetitive regions was greater in non-cancerous tissue and plasma DNA of healthy subjects when compared to tumor tissue.

  As a result, plasma DNA of cancer patients showed a greater reduction in methylation density in the repeat region than in the non-repetitive region. The difference in plasma DNA methylation density between the mean of 4 healthy controls and between HCC patients was 0.163 and 0.088 in the repetitive and non-repetitive regions, respectively. The data for pre- and post-operative plasma samples also showed that the dynamic range of changes in methylation density was greater in the repetitive region than in the non-repetitive region. In one embodiment, plasma DNA methylation densities of repetitive regions can be used to determine whether a patient has cancer or monitor disease progression.

  As mentioned above, variations in the methylation density in the reference plasma also affect the accuracy with which cancer patients are distinguished from non-cancer individuals. The closer the distribution of methylation densities (ie, the smaller the standard deviation), the more accurate the distinction between cancer and non-cancer subjects. In another embodiment, the coefficient of variation of methylation density (CV) of 1 Mb bin can be used as a criterion for selecting bins with low variability of plasma DNA methylation density in the reference group. For example, only bins with a CV of less than 1% are selected. Other values, such as 0.5%, 0.75%, 1.25% and 1.5% can also be used as criteria for selecting bins with low methylation density variability. In yet another embodiment, the selection criteria may include both the CV of the bin and the difference in methylation density between cancerous and non-cancerous tissues.

  Methylation density can also be used to estimate the concentration fraction of tumor-derived DNA in plasma samples where the methylation density of the tumor tissue is known. This information can be obtained by analysis of the patient's tumor or from examination of tumors obtained from several patients with the same cancer type. As mentioned above, the plasma methylation density (P) can be expressed using the following equation:

Where BKG is the background methylation density obtained from blood cells and other organs, TUM is the methylation density in tumor tissue, and f is the concentration fraction of tumor-derived DNA in plasma samples. This can be rewritten as:

  BKG values can be determined by analyzing plasma samples of patients at the absence of cancer or from a survey of a reference group of individuals without cancer. Therefore, f can be determined after measuring the plasma methylation density.

F. Combination with other methods The methylation analysis approach described herein can be used in combination with other methods based on genetic alterations of tumor-derived DNA in plasma. Examples of such methods include cancer-associated chromosomal abnormalities (KCA Chan et al. 2013 Clin Chem; 59: 211-224; RJ Leary et al. 2012 Sci Transl Med; 4: 162ra154) and cancer-related in plasma. Analysis of single nucleotide mutations (KCA Chan et al. 2013 Clin Chem; 59: 211-224) is included. The methylation analysis approaches have advantages over their genetic approaches.

  As shown in FIG. 21A, hypomethylation of tumor DNA is a global phenomenon that includes regions distributed almost throughout the genome. Therefore, DNA fragments from all chromosomal regions are beneficial with respect to the potential contribution of tumor-derived hypomethylated DNA to plasma / serum DNA in patients. In contrast, chromosomal abnormalities (amplification or deletion of chromosomal regions) are only present in some chromosomal regions, and DNA fragments from regions without chromosomal abnormalities in tumor tissue are not informative (KCA). Chan et al. 2013 Clin Chem; 59: 211-224). Similarly, only thousands of single nucleotide changes are observed in each cancer genome (KCA Chan et al. 2013 Clin Chem; 59: 211-224). DNA fragments that do not overlap with these single nucleotide changes are not useful in determining whether tumor-derived DNA is present in plasma. Thus, the present methylation analysis approaches are potentially more cost-effective than their genetic approaches to detect cancer-related changes in circulating blood.

  In one embodiment, the cost-effectiveness of plasma DNA methylation analysis is due to the DNA fragments from the most informative regions, such as the most highly differential methylation differences between cancerous and non-cancerous tissues. Can be further enhanced by enriching the region with Examples of methods for enriching these regions include the use of hybridization probes (eg, Nimburgen SeqCap system and Agilent SureSelect Target Enrichment system), PCR amplification and solid phase hybridization.

G. Tissue-specific analysis / donor Tumor-derived cells invade and metastasize to adjacent or distant organs. Infiltrated tissues or metastases supply DNA into the plasma as a result of cell death. By analyzing the methylation characteristics of DNA in plasma of cancer patients and detecting the presence of a tissue-specific methylation signature, the tissue type involved in the disease process can be detected. This approach helps identify organs involved as primary and metastatic sites by providing a non-invasive anatomical scan of tissues involved in the cancer process. In addition, by monitoring the relative levels of methylation signatures in diseased organs in plasma, tumor burden in those organs can be assessed and the cancer process in those organs is worsening or improving or the cure is complete. It becomes possible to decide whether or not. For example, suppose gene X is specifically methylated in the liver. In that case, metastatic involvement of the liver by the cancer (eg, colorectal cancer) would be expected to increase the concentration of the methylated sequence from gene X in plasma. There may be other sequences or sequences with similar methylation characteristics as gene X. In that case, the results obtained from such sequences can be combined. Similar considerations apply to other tissues, such as brain, bone, lungs and kidneys.

  On the other hand, DNAs derived from different organs are known to show a tissue-specific methylation signature (BW Futscher et al. 2002 Nat Genet; 31: 175-179; SSC Chim et al. 2008 Clin Chem; 54: 500. -511). Therefore, plasma methylation profiling can be used to elucidate the contribution of tissues from various organs to plasma. Since plasma DNA is believed to be released when cells die, elucidation of such contributions can be used to assess organ injury. For example, liver lesions such as hepatitis (eg, due to viruses, autoimmune processes, etc.) or hepatotoxicity (eg, drug overdose (eg, paracetamol) or toxins (alcohol, etc.) caused by drugs are associated with hepatocyte damage. It is related and is expected to be associated with an increase in the level of liver-derived DNA in plasma, eg, gene X is specifically methylated in the liver, in which case liver lesions are , It is expected to increase the concentration of methylated sequences derived from gene X in plasma. Conversely, gene Y is specifically hypomethylated in the liver, in which case liver lesions It is expected to reduce the concentration of methylated sequences from gene Y. In yet other embodiments, gene X or gene Y may be different from the body, which may not be the gene. It may be replaced by any genomic sequence show differential methylation in weaving.

  The method described herein can also be applied to the evaluation of donor-derived DNA in plasma of organ transplant recipients (YMD Lo et al. 1998 Lancet; 351: 1329-1330). Polymorphic differences between donors and recipients were used to distinguish donor-derived DNA from recipient-derived DNA in plasma (YW Zheng et al. 2012 Clin Chem; 58: 549-558). We report that the tissue-specific methylation signature of transplanted organs can also be used as a method to detect donor DNA in the plasma of recipients.

  By monitoring the concentration of the donor DNA, the condition of the transplanted organ can be evaluated non-invasively. For example, transplant rejection is associated with higher cell death rates, and thereby higher concentrations of donor DNA in recipient plasma (or serum) as reflected by the methylation signature of transplanted organs, Would be increased when compared to the stable state or to other stable transplant recipients or non-transplanted healthy controls. Similar to what has been reported for cancer, donor-derived DNA is polymorphic difference, shorter size DNA of transplanted parenchymal organs (YW Zheng et al. 2012 Clin Chem; 58: 549-558). And by detecting all or some of the properties, including tissue-specific methylation properties, can be identified in the plasma of transplant recipients.

H. Normalization of Methylation Based on Size As described above and in Lun et al (FMF Lun et al. Clin. Chem. 2013; doi: 10.1373 / clinchem.2013.212274), methylation density (eg plasma DNA Methylation density) correlates with the size of the DNA fragment. The distribution of methylation densities of shorter plasma DNA fragments was significantly lower than that of longer fragments. Some non-cancerous conditions (eg, systemic lupus erythematosus (SLE)) with an abnormal pattern of plasma DNA fragmentation are apparent due to the presence of larger amounts of shorter plasma DNA fragments that are more hypomethylated. We propose that it may exhibit plasma DNA hypomethylation. In other words, the size distribution of plasma DNA can be a confounder of the methylation density of plasma DNA.

  FIG. 34A shows the size distribution of plasma DNA in SLE patient SLE04. The size distribution of 9 healthy control patients is shown as the gray dotted line and the size distribution of SLE04 is shown as the black solid line. Short plasma DNA fragments were more abundant in SLE04 than in 9 healthy control patients. This size distribution pattern can disrupt the methylation analysis of plasma DNA, resulting in additional apparent hypomethylation, as shorter DNA fragments are generally more hypomethylated.

  In some embodiments, the measured methylation levels can be normalized to reduce the confounding effect of size distribution on plasma DNA methylation analysis. For example, the size of the DNA molecule at multiple sites can be measured. In various implementations, measurements can simply determine that a size is within a particular range, which can, or can correspond to, a particular size (eg, length) of a DNA molecule. The normalized methylation level can then be compared to the cutoff value. There are several ways to perform normalization to reduce confounding effects on plasma DNA methylation analysis of size distribution.

  In one embodiment, size fractionation of DNA (eg, plasma DNA) can be performed. Size fractionation may ensure that DNA fragments with similar sizes are used to determine methylation levels in a manner consistent with cutoff values. As part of the size fraction, a DNA fragment having a first size (eg, a first length range) can be selected, where the first cutoff value corresponds to the first size. . Normalization can be achieved by calculating methylation levels using only those selected DNA fragments.

  Size fractionation can be performed in various ways, for example by physical separation of DNA molecules of different sizes (for example by electrophoretic or microfluidics-based technology or by centrifugation-based technology) or by in silico analysis. Can be achieved. In in silico analysis, in one embodiment, paired-end massively parallel sequencing of plasma DNA molecules can be performed. The size of the sequenced molecule can then be estimated by comparing the position of each of the two ends of the plasma DNA molecule to the reference human genome. Subsequent analysis can then be performed by selecting sequenced DNA molecules that meet one or more size selection criteria (eg, a size criterion within a particular range). Thus, in one embodiment, the methylation density of fragments with smaller size (eg, within a certain range) can be analyzed. The cutoff value (eg, at block 2840 of method 2800) may be determined based on the fragments within the same size range. For example, methylation levels can be determined from samples known to have cancer or not, and cutoff values can be determined from these methylation levels.

  In another embodiment, a functional relationship between methylation density and size of circulating DNA can be determined. Functional relationships can be defined by the data points or coefficients of the function. The functional relationship may provide a scaling value corresponding to each size (eg, shorter size may have a corresponding increase for methylation). In various implementations, the scaling value can be 0-1 or greater than 1.

  Normalization can be done based on average size. For example, the average size corresponding to the DNA molecules used to calculate the first methylation level can be computed and the first methylation level corresponds to a corresponding scaling value (ie, corresponding to the average size). Can be multiplied by. As another example, the methylation density of each DNA molecule can be normalized according to the size of the DNA molecule and the relationship between DNA size and methylation.

  In another implementation, normalization can be done on a molecule-by-molecule basis. For example, the size of each of the DNA molecules at a particular site can be obtained (eg, as described above) and the scaling value corresponding to each size can be identified from the functional relationship. In the non-normalized calculation, each molecule is counted equally in determining the methylation index at the site. In a normalized calculation, the contribution of a molecule to the methylation index can be weighted by a scaling factor that corresponds to the size of the molecule.

34B and 34C show methylation analysis of plasma DNA from SLE patient SLE04 (FIG. 34B) and HCC patient TBR36 (FIG. 34C). The outer circle shows the Z _meth result for plasma DNA without in silico size fractionation. The inner ring shows the Z _meth result for plasma DNA above 130 bp. In SLE patient SLE04, 84% of the bins showed hypomethylation without in silico size fractionation. The percentage of bins exhibiting hypomethylation was reduced to 15% when only fragments of 130 bp and above were analyzed. In HCC patient TBR36, 98.5% and 98.6% of the bins showed hypomethylation of plasma DNA with or without in silico size fractionation, respectively. These results indicate that in silico size fractionation effectively reduces false positive hypomethylation results associated with increased fragmentation of plasma DNA (eg, in patients with SLE or other inflammatory diseases). Suggests to get.

  In one embodiment, the results of analyzes with and without size fractionation can be compared to indicate if there is a size confounding effect on methylation results. Therefore, in addition to, or instead of, normalization, the calculation of methylation levels at a particular size may result in false positives when the percentage of bins above the cutoff value differs with and without size fractionation. Can be used to determine whether or not a particular methylation level differs. For example, the presence of a significant difference between the results of samples with and without size fractionation can be used to indicate the likelihood of false positive results due to aberrant fragmentation patterns. Thresholds for determining whether a difference is significant can be established by analysis of cohorts of cancer patients and non-cancer control patients.

I. Analysis of genome-wide CG island hypermethylation in plasma In addition to global hypomethylation, CG island hypermethylation is commonly observed in cancer (SSB Baylin et al. 2011 Nat Rev Cancer ; 11: 726-734; PA Jones et al. 2007, Cell; 128: 683-692; M Esteller et al. 2007 Nat Rev Genet 2007; 8: 286-298; M Ehrlich et al. 2002 Oncogene 2002; 21: 5400-5413). This section describes the use of genome-wide analysis of CG island hypermethylation to detect and monitor cancer.

  FIG. 35 is a flowchart of a method 3500 for determining a cancer level classification based on CG island hypermethylation, according to an embodiment of the invention. The plurality of sites of method 2800 can include CpG sites, where the CpG sites are assembled into multiple CG islands, each CG island including one or more CpG sites. The methylation level of each CG island can be used to determine the classification of cancer levels.

  At block 3510, CG islands to be analyzed are identified. In this analysis, as an example, a series of CG islands to be analyzed were defined, which were characterized by a relatively low methylation density in plasma of a healthy reference. In one aspect, the change in methylation density in the reference group can be relatively small to allow for easier detection of cancer-associated hypermethylation. In one embodiment, the CG islands have an average methylation density that is less than the first percentage in the reference group and the coefficient of variation of methylation density in the reference group is less than the second percentage.

As an example, for illustrative purposes, the following criteria are used to identify useful CG islands:
i. Average methylation density of CG islands in the reference group (eg healthy subjects) is less than 5% ii. Coefficient of variation of analysis of methylation density in plasma of reference group (eg healthy subjects) is less than 30%. These parameters can be adjusted for the particular application. From our dataset, 454 CG islands within the genome met these criteria.

  At block 3520, the methylation density for each CG island is calculated. Methylation density can be calculated as described herein.

  At block 3530, it is determined whether each CG island is hypermethylated. For example, in the analysis of CG island hypermethylation of the tested cases, the methylation density of each CG island was compared to the corresponding data of the reference group. Methylation density (an example of methylation level) can be compared to one or more cutoff values to determine whether a particular island is hypermethylated.

In one embodiment, the first cutoff value may correspond to the mean value of the methylation density of the reference group + a specific percentage. Another cut-off value may correspond to the mean of the methylation densities of the reference group + a specific standard deviation value. In one implementation, the Z value (Z _meth ) was calculated and compared to the cutoff value. As an example, a CG island in a test subject (eg, a subject who has been screened for cancer) was considered to be significantly hypermethylated if the following criteria were met:
i. Its methylation density is more than 2% higher than the mean value of the reference group,
ii. Z _meth is greater than 3.
Also, these parameters can be adjusted for the particular application.

At block 3540, the methylation density (eg, Z value) of hypermethylated CG islands is used to determine a cumulative score. For example, after identifying all significantly hypermethylated CG islands, a score can be calculated that includes the sum of Z values or a function of the Z values of all hypermethylated CG islands. An example of a score is the cumulative probability (CP) score described in another section. Cumulative probability scores use Z _meth to determine the probability of having such observations by chance, according to a probability distribution (eg, Student's t probability distribution with 3 degrees of freedom).

  At block 3550, the cumulative score is compared to a cumulative threshold to determine the level classification of the cancer. For example, an organism can be identified as having cancer if the total hypermethylation at the identified CG island is sufficiently high. In one embodiment, the cumulative threshold matches the highest cumulative score obtained from the reference group.

IX. Methylation and CNA
As mentioned above, the methylation analysis approach described herein can be used in combination with other methods based on genetic alterations of tumor-derived DNA in plasma. Examples of such methods include analysis of cancer-associated chromosomal abnormalities (KCA Chan et al. 2013 Clin Chem; 59: 211-224; RJ Leary et al. 2012 Sci Transl Med; 4: 162ra154). Aspects of Copy Number Abnormality (CNA) are described in US Patent Application Publication No. 13 / 308,473.

A. CNA
Copy number abnormalities can be detected by counting the DNA fragments that align with a particular portion of the genome, normalizing the count and comparing the count to a cutoff value. In various embodiments, normalization is performed by counting DNA fragments that are aligned with different haplotypes of the same portion of the genome (relative haplotype dosage (RHDO)) or with DNA fragments that are aligned with other portions of the genome. Can be done by counting.

  The RHDO method relies on the use of heterozygous loci. The embodiments described in this section can also be used for homozygous loci by comparing two regions and not two haplotypes of the same region, and thus are non-haplotype specific. . In the relative chromosomal region dosage method, the number of fragments from one chromosomal region (eg, determined by counting sequence reads aligned to that region) is the expected value (reference chromosome). Region, or from the same region of another sample known to be healthy). Thus, fragments of chromosomal regions are counted regardless of which haplotype the sequenced tag came from. Therefore, sequence reads containing non-heterozygous loci can also be used. To make the comparison, in some embodiments, the tag count may be normalized before the comparison. Each region is defined by at least two loci (separated from each other) and the fragments at these loci can be used to obtain a collective value for that region.

  The normalized number of sequenced reads (tags) in a particular region is the number of sequenced reads aligned in that region divided by the total number of sequenced reads that can be aligned throughout the genome. Can be calculated by This normalized tag number allows the results obtained from one sample to be compared with the results of another sample. For example, the normalization number can be the ratio (eg, percentage or fraction) of sequenced reads expected to be from a particular region, as described above. In other embodiments, other methods for normalization are possible. For example, one can normalize by dividing the number of counts in one region by the number of counts in the reference region (in the above example, the reference region is simply the entire genome). This normalized tag number can then be compared to a threshold that can be determined from one or more reference samples that do not show cancer.

  The normalized tag number of the test case is then compared to the normalized tag number of one or more reference subjects (eg, one or more reference subjects without cancer). In one embodiment, the comparison is performed by calculating the Z value of the case in a particular chromosomal region. The Z-value can be calculated using the following formula: Z-value = (normalized tag number of cases-mean) / SD, where "mean" aligns with a particular chromosomal region of the reference sample. Is the average of the normalized tag numbers; SD is the standard deviation of the number of normalized tag numbers aligned in a particular region of the reference sample. Therefore, the Z value is the number of standard deviations in which the normalized number of tags of the chromosomal region of the examined case deviates from the normalized average number of tags of the same chromosomal region of one or more reference objects. is there.

  In situations where the test organism has cancer, the chromosomal regions amplified in tumor tissue make up a large proportion of plasma DNA. This results in a positive Z value. On the other hand, chromosomal regions deleted in tumor tissue occupy a small proportion in plasma DNA. This results in a negative Z value. The magnitude of the Z value is determined by several factors.

  One factor is the concentration fraction of tumor-derived DNA in a biological sample (eg plasma). The higher the concentration fraction of tumor-derived DNA in the sample (eg plasma), the greater the difference between the normalized tag numbers of the test and reference cases. Therefore, a larger magnitude Z value is provided.

  Another factor is the variation in normalized tag number in one or more reference cases. If the over-presentation of chromosomal regions in the biological samples of the tested cases (eg plasma) is of similar extent, smaller variations in the normalized tag number in the reference group (ie smaller standard deviations) lead to larger Z values. Bring Similarly, a smaller standard deviation of the normalized tag number in the reference group leads to a more negative Z-value if the underpresentation of chromosomal regions in the biological samples (eg plasma) of the tested cases is of similar extent.

  Another factor is the magnitude of chromosomal aberrations in tumor tissue. The magnitude of a chromosomal abnormality refers to a copy number change (increase or decrease) of a specific chromosomal region. The higher the copy number change in tumor tissue, the higher the degree of over- or under-presentation of a particular chromosomal region in plasma DNA. For example, loss of both copies of a chromosome results in greater under-representation of chromosomal regions in plasma DNA than loss of one of the two copies of the chromosome, and thus an even more negative Z value. Typically, cancer has multiple chromosomal abnormalities. Chromosome abnormalities in each cancer further depend on their nature (ie, amplification or deletion), their extent (increasing or decreasing single or multiple copies) and their extent (size of the abnormality with respect to chromosome length). obtain.

  The accuracy of measuring the normalized number of tags is affected by the number of molecules analyzed. To detect chromosomal abnormalities with one copy change (increase or decrease), 15,000, 60,000 when the concentration fractions are approximately 12.5%, 6.3% and 3.2%, respectively. And 240,000 molecules are expected to need to be analyzed. Further details of tag counting in the detection of cancer in different chromosomal regions is described in Lo et al., US Patent Application Publication No. 2009/0029377, entitled "Diagnosing Fetal Chromosomal Aneuploidy Using Massively Parallel Genomic Sequencing", The entire contents of the US patent application publication are hereby incorporated by reference for all purposes.

  In embodiments, size analysis may also be used instead of the tag counting method. Size analysis can also be used instead of normalized tag number. Various parameters described herein and in US Patent Application Publication No. 12 / 940,992 may be used in size analysis. For example, the Q or F value from above can be used. Such size values do not require normalization by counting from other areas, as these values do not correspond to the number of reads. Haplotype-specific method approaches, such as the RHDO method described above and described in more detail in US Patent Application Publication No. 13 / 308,473, can also be used for non-specific methods. For example, techniques involving region depth and refinement may be used. In some embodiments, the GC trend of a particular region may be considered when comparing the two regions. Since the RHDO method uses the same area, such correction is unnecessary.

  Although some cancers may typically be present with abnormalities in particular chromosomal regions, such cancers are always exclusively present with abnormalities in such regions. It doesn't mean that. For example, additional chromosomal regions may exhibit abnormalities and the location of such additional regions may be unknown. Furthermore, when screening patients to identify early stages of cancer, it may be desirable to identify a wide range of cancers that may exhibit abnormalities that exist throughout the genome. To address these situations, embodiments may systematically analyze multiple regions to determine which regions exhibit anomalies. The number of abnormalities and their location (eg, whether they are in close proximity), for example, identify the abnormalities, determine the stage of the cancer, and make a cancer diagnosis (eg, the number is above a threshold). If too large), it can be used to make a prognosis based on the number and location of various regions exhibiting abnormalities.

  Therefore, in an embodiment, whether or not an organism has cancer can be identified based on the number of regions that show an abnormality. Therefore, multiple regions (eg, 3,000) may be examined to identify the number of regions that exhibit anomalies. The region may span the entire genome or may span only a portion of the genome (eg, a non-repetitive region).

  FIG. 36 is a flow chart of a method 3600 for analyzing a biological sample of an organism using multiple chromosomal regions according to an embodiment of the present invention. Biological samples include nucleic acid molecules (also called fragments).

  At block 3610, multiple regions (eg, non-overlapping regions) of the organism's genome are identified. Each chromosomal region contains multiple loci. The region may be 1 Mb in size, or some other comparable size. If the region is 1 Mb in size, the entire genome may include approximately 3,000 regions each having a given size and position. Such a given region may vary to accommodate a particular chromosome length or a particular number of regions used, and any other criteria described herein. If the regions have different lengths, such lengths can be used to normalize the results, eg, as described herein. Regions may be specifically selected based on certain criteria for a particular organism and / or based on the findings of the cancer being tested. Areas can also be arbitrarily selected.

  At block 3620, the location of the nucleic acid molecule within the organism's reference genome is identified for each of the plurality of nucleic acid molecules. The location can be determined by any of the methods described herein, for example, by sequencing the fragment to obtain a sequenced tag and aligning the sequenced tag with a reference genome. . The particular haplotype of the molecule can also be determined for haplotype specific methods.

  Blocks 3630-3650 are performed for each of the chromosomal regions. At block 3630, each nucleic acid molecule group is identified as originating from the chromosomal region based on the identified location. Each group may include at least one nucleic acid molecule located at each of a plurality of loci in the chromosomal region. In one embodiment, the group may be fragments that align with a particular haplotype of the chromosomal region (eg, in RHDO as described above). In another embodiment, the group can be any fragment that aligns with the chromosomal region.

  At block 3640, the computer system calculates each value for each nucleic acid molecule population. Each value defines the identity of each group of nucleic acid molecules. Each value can be any of the values described herein. For example, the value can be a statistic of the number of fragments in a group or the size distribution of fragments in a group. Each value can also be a normalized number, eg, the total number of tags in the sample or the number of tags in the region divided by the number of tags in the reference region. Each value may be a difference or ratio from another value (eg, in RHDO), which may give rise to a characteristic of the difference in area.

  At block 3650, each value is compared to a reference value to determine a classification as to whether the first chromosomal region exhibits a deletion or amplification. This reference value may be any threshold value or reference value described herein. For example, the reference value can be a threshold value determined in a normal sample. In RHDO, each value can be the difference or ratio of tag numbers in the two haplotypes, and the reference value can be a threshold for determining that a statistically significant deviation exists. As another example, the reference value can be the tag number or size value of another haplotype or region, and the comparison can include obtaining a difference or ratio (or a function of such ratio), and then the difference or ratio being a threshold value. May be included.

  The reference value may vary based on the results of other areas. For example, if adjacent regions also exhibit deviations (though small compared to some threshold (eg, a Z value of 3)), a smaller one may be used. For example, if all three consecutive regions are above the first threshold, then it is likely a cancer. Therefore, this first threshold may be lower than another threshold needed to identify cancer from non-contiguous regions. Having three regions (or more than three) with at least deviations can be sufficiently low in chance of chance effects where sensitivity and specificity can be preserved.

  At block 3660, the amount of genomic region classified as exhibiting a deletion or amplification is determined. There may be restrictions on the chromosomal regions that are counted. For example, only areas that are adjacent to at least one other area may be counted (or the adjacent areas may be required to be of a certain size (eg, 4 or more areas)). In embodiments where the regions are not equal, the numbers may also account for their respective lengths (eg, the numbers may be the total length of the abnormal region).

  At block 3670, the quantity is compared to a quantity threshold to determine a sample classification. As an example, the classification can be whether the organism has cancer, the stage of the cancer, and the prognosis of the cancer. In one embodiment, all abnormal areas are counted and a single threshold is used regardless of where those areas appear. In another embodiment, the threshold may vary based on the position and size of the counted area. For example, the amount of a region on a particular chromosome or arm can be compared to a threshold for that particular chromosome (or arm). Multiple thresholds may be used. For example, the amount of abnormal regions on a particular chromosome (or arm) must be greater than the first threshold, and the total amount of abnormal regions in the genome must be greater than the second threshold. I won't. The threshold may be the percentage of the area determined to show deletion or amplification.

  This threshold for the amount of area may also depend on how strong the imbalance is in the counted area. For example, the amount of region used as a threshold to determine the classification of cancer may depend on the specificity and sensitivity (abnormal threshold) used to detect anomalies in each region. For example, if the abnormal threshold is low (eg, a Z value of 2), the quantity threshold may be selected to be high (eg, 150). However, if the abnormal threshold is high (eg a Z value of 3), the quantity threshold may be low (eg 50). Also, the amount of regions exhibiting anomalies may be weighted, for example, one region exhibiting a high degree of imbalance may be weighted higher than a region exhibiting only a small imbalance (ie, simply positive for anomalies and There are more classifications than negatives). As an example, the sum of the Z values can be used, and thus the weighted value is used.

  Therefore, the amount of chromosomal regions (which may include number and / or size) that show significant over- or under-presentation of the normalized tag number (or each other value in the group's characteristics) reflects the severity of the disease. Can be used to The amount of chromosomal regions with aberrant normalized tag numbers is determined by two factors: the number (or size) of chromosomal abnormalities within the tumor tissue and the concentration fraction of tumor-derived DNA in the biological sample (eg plasma). Can be determined by More advanced cancers tend to show more (and larger) chromosomal abnormalities. Therefore, more cancer-related chromosomal abnormalities may be detectable in a sample (eg plasma). In patients with more advanced cancer, higher tumor burden will result in a higher concentration fraction of tumor-derived DNA in plasma. As a result, tumor-associated chromosomal abnormalities will be more easily detected in plasma samples.

  One possible approach to improve sensitivity without sacrificing specificity is to consider the results of adjacent chromosomal segments. In one embodiment, the Z-value cutoff remains above 2 and below -2. However, a chromosomal region is classified as potentially abnormal only if two consecutive segments exhibit the same type of abnormality (eg, both segments have Z values greater than 2). In other embodiments, the Z-values of adjacent segments may be summed with higher cutoff values. For example, the Z values of three consecutive segments may be summed and a cutoff value of 5 may be used. This concept can be extended to more than three consecutive segments.

  The combination of quantity and abnormal threshold may also depend on the purpose of the analysis and any prior knowledge (or lack thereof) of the organism. For example, if a healthy population is screened for cancer, typically both at the amount of area (ie, a high threshold for the number of areas) and possibly at an abnormal threshold when the area is identified as having abnormalities, High specificity is used. However, patients at higher risk (eg, patients with lump or family history complaints, smokers, chronic human papillomavirus (HPV) carriers, hepatitis virus carriers, or other viral carriers) ), The threshold can be lower in order to have higher sensitivity (less false negatives).

  In one embodiment, if a lower detection limit of 1 Mb resolution and 6.3% tumor-derived DNA is used to detect chromosomal abnormalities, then the number of molecules in each 1 Mb segment must be 60,000. Becomes This replaces approximately 180 million (60,000 reads / Mb x 3,000 Mb) alignable reads throughout the genome.

  The smaller segment size gives higher resolution for detecting smaller chromosomal abnormalities. However, this increases the requirement for the number of molecules analyzed in total. Larger segment sizes reduce the number of molecules needed for analysis at the expense of resolution. Therefore, only larger anomalies can be detected. In some implementations, larger regions may be used, segments that show anomalies may be subdivided, and these small regions may be analyzed for better resolution (eg, as described above). Having a conjecture on the size of the deletion or amplification detected (or the lowest concentration to detect), the number of molecules analyzed can be reduced.

B. CNA based on sequencing of bisulfite treated plasma DNA
Genome-wide hypomethylation and CNAs can frequently be observed in tumor tissue. Here we show that information on CNA and cancer-related methylation changes can be obtained simultaneously from bisulfite sequencing of plasma DNA. Since the two types of analysis can be performed on the same dataset, there is virtually no additional cost for CNA analysis. In other embodiments, different procedures may be used to obtain methylation and genetic information. In other embodiments, a similar analysis for cancer associated hypermethylation may be performed in combination with CNA analysis.

  FIG. 37A shows a CNA analysis (inside to outside) of tumor tissue, non-bisulfite (BS) -treated and bisulfite-treated plasma DNA of patient TBR36. FIG. 37A shows a CNA analysis (inside to outside) of tumor tissue, non-bisulfite (BS) -treated and bisulfite-treated plasma DNA of patient TBR36. The outermost ring shows a schematic diagram of the chromosome. Each dot represents the result of the 1 Mb area. Green, red and gray dots represent areas with increased copy number, areas with decreased copy number and areas with no changed copy number, respectively. Z values are shown in plasma analysis. There are 5 differences between the two concentric lines. Tumor tissue analysis indicates copy number. There is one copy difference between the two concentric lines. FIG. 38A shows a CNA analysis of patient TBR34 tumor tissue, non-bisulfite (BS) treated plasma DNA and bisulfite treated plasma DNA (from inside to outside). The patterns of CNA detected in the bisulfite-treated and non-bisulfite-treated plasma samples were in agreement.

  The patterns of CNA detected in tumor tissue, non-bisulfite treated plasma and bisulfite treated plasma were consistent. Scatter plots are generated to further evaluate the agreement between the results for the bisulfite-treated and non-bisulfite-treated plasmas. FIG. 37B is a scatter plot showing the relationship between Z values in the detection of CNA using 1 Mb bin bisulfite-treated plasma and non-bisulfite-treated plasma of patient TBR36. A positive correlation was observed between the Z values of those two analyzes (r = 0.89, p <0.001, Pearson correlation). FIG. 38B is a scatter plot showing the association between Z values in the detection of CNA using 1 Mb bin bisulfite-treated and non-bisulfite-treated plasma of patient TBR34. A positive correlation was observed between the Z values of those two analyzes (r = 0.81, p <0.001, Pearson correlation).

C. Synergistic Analysis of Cancer-Associated CNAs and Methylation Changes As noted above, analysis of CNAs may involve counting the number of sequence reads within each 1 Mb region, while analysis of methylation density indicates that methylation is Detection of the percentage of cytosine residues in the existing CpG dinucleotide can be included. The combination of these two analyzes can provide synergistic information in cancer detection. For example, the methylation classification and CNA classification can be used to determine a third classification of cancer levels.

  In one embodiment, the presence of either cancer-associated CNAs or methylation changes may be used to indicate the potential presence of cancer. In such embodiments, the sensitivity of detecting cancer may be increased when CNA or methylation changes are present in the plasma under test. In another embodiment, the presence of both changes can be used to indicate the presence of cancer. In such an embodiment, the specificity of the test can be increased because either of these two types of changes can potentially be detected in some non-cancerous subjects. Thus, the third classification can be cancer positive only if both the first and second classifications indicate cancer.

  Twenty-six HCC patients and 22 healthy subjects were recruited. Blood samples were taken from each subject and plasma DNA was sequenced after bisulfite treatment. In HCC patients, blood samples were taken at diagnosis. The presence of a significant amount of CNA was defined as having bins greater than 5% with, for example, Z values less than -3 or greater than 3. The presence of a significant amount of cancer-associated hypomethylation was defined as having more than 3% bins with Z values less than -3. As an example, the amount of area (bin) may be expressed as the raw number of bins, the percentage, and the length of the bin.

  Table 3 shows the detection of significant amounts of CNA and methylation changes in plasma of 26 HCC patients using massively parallel sequencing on bisulfite treated plasma DNA.

  The detection rates of cancer-related methylation changes and CNAs were 69% and 50%, respectively. When either criterion's presence was used to indicate the potential presence of cancer, the detection rate (ie, diagnostic sensitivity) improved to 73%.

  Results from two patients showing the presence of CNA (Figure 39A) or methylation changes (Figure 39B) are shown. FIG. 39A is a Circos plot showing CNA (inner ring) and methylation analysis (outer ring) of bisulfite treated plasma of HCC patient TBR240. In CNA analysis, green, red and gray dots represent regions with increased chromosomes, decreased chromosomes and unchanged copy number, respectively. In the methylation analysis, green intent, red and gray dots represent regions with hypermethylation, hypomethylation and normal methylation, respectively. Cancer-associated CNAs were detected in plasma in this patient, but methylation analysis did not show significant amounts of cancer-associated hypomethylation. FIG. 39B is a Circos plot showing CNA (inner ring) and methylation analysis (outer ring) of bisulfite treated plasma of HCC patient TBR164. Cancer-related hypomethylation was detected in plasma in this patient. However, no significant amount of CNA could be observed. The results of two patients showing the presence of both CNA and methylation changes are shown in Figure 48A (TBR36) and Figure 49A (TBR34).

  Table 4 shows the detection of significant amounts of CNA and methylation changes in the plasma of 22 control patients using massively parallel sequencing on bisulfite treated plasma DNA. The bootstrapping (ie, stripping) method was used to evaluate each control patient. Therefore, when a particular subject was evaluated, the other 21 subjects were used to calculate the mean and SD of the control group.

  The specificity of detection of significant amounts of methylation changes and CNA was 86% and 91%, respectively. Specificity increased to 95% when the presence of both criteria was required to indicate the potential presence of cancer.

  In one embodiment, CNA and / or hypomethylation positive samples are considered cancer positive and samples are considered negative if both are undetectable. Higher sensitivity is provided by using the "or" reasoning. In another embodiment, only samples that are positive for both CNA and hypomethylation are considered cancer positive, thereby conferring higher specificity. In yet another embodiment, a three tier classification may be used. The subject is i. Both are normal; ii. One is abnormal; iii. Both are classified as abnormal.

  Different tracking strategies can be used for these three classifications. For example, the subject in (iii) may be subjected to the most intensive tracking protocol (including, eg, whole body imaging); the subject in (ii) may be less intensive in the tracking protocol (eg, relatively short, such as weeks). Repetitive plasma DNA sequencing after time intervals) may be administered; subjects in (i) may be subjected to the least intensive follow-up protocol (eg, retest at some years later). In other embodiments, methylation and CNA measurements can be used in combination with other clinical parameters to further subdivide the classification, such as imaging results or serum biochemistry.

D. Prognostic value of plasma DNA analysis after therapeutic treatments The presence of cancer-associated CNA and / or methylation changes in plasma indicates the presence of tumor-derived DNA in the circulating blood of cancer patients. Reduction or elimination of these cancer-related changes is expected after treatment (eg, surgery). On the other hand, the persistence of these changes in plasma after treatment may indicate incomplete removal of all tumor cells from the body and may be a useful prognostic factor for disease recurrence.

  Blood samples were taken from two HCC patients, TBR34 and TBR36, one week after surgical resection aimed at treating the tumor. CNA and methylation analyzes were performed on bisulfite treated post-treatment plasma samples.

  FIG. 40A shows CNA analysis on bisulfite treated plasma DNA taken before (inner ring) and after (outer ring) surgical resection of tumor in HCC patient TBR36. Each dot represents the result of the 1 Mb area. The green, red and gray dots represent the areas with increased copy number, the areas with decreased copy number and the areas with unchanged copy number, respectively. Most of the CNA observed before treatment disappeared after tumor resection. The percentage of bins with Z-values below -3 or above 3 decreased from 25% to 6.6%.

  FIG. 40B shows methylation analysis on bisulfite treated plasma DNA taken before (inner ring) and after (outer ring) surgical resection of tumors in HCC patient TBR36. Green, red and gray dots represent regions with hypermethylation, hypomethylation and normal methylation, respectively. A significant decrease in the percentage of bins showing significant hypomethylation was seen from 90% to 7.9%, and the degree of hypomethylation also showed a significant decrease. This patient had a complete clinical remission 22 months after tumor resection.

  FIG. 41A shows a CNA analysis on bisulfite treated plasma DNA taken before (inner ring) and after (outer ring) surgical resection of tumor in HCC patient TBR34. Both the number of CNA-representing bins and the magnitude of CNA in the affected bins after surgical resection of the tumor are reduced, but residual CNA can be observed in post-operative plasma samples. The red ring highlights the area where residual CNA was most apparent. The percentage of bins with Z-values below -3 or above -3 was reduced from 57% to 12%.

  FIG. 41B shows methylation analysis on bisulfite treated plasma DNA taken before (inner ring) and after (outer ring) surgical resection of tumor in HCC patient TBR34. The magnitude of hypomethylation decreased after tumor resection, and the mean Z value for hypomethylated bins decreased from -7.9 to -4.0. However, the percentage of bins with Z values less than -3 showed the opposite change, increasing from 41% to 85%. This observation potentially indicates the presence of residual cancer cells after treatment. Clinically, multiple lesions of tumor nodules were detected in the remaining non-resected liver 3 months after tumor resection. Lung metastases were observed 4 months after surgery. The patient died 8 months after surgery due to local recurrence and metastatic disease.

  Observations in these two patients (TBR34 and TBR36) indicate that the presence of residual cancer-related changes in CNA and hypomethylation should be used for the monitoring and prognosis of post-treatment cancer patients for therapeutic purposes. Suggest that you can. The data also show that the degree of change in the amount of plasma CNA detected can be used synergistically with the assessment of the degree of change in the range of plasma DNA hypomethylation for prognostic and monitoring therapeutic efficacy. I showed that.

  Thus, in some embodiments, one biological sample is obtained before treatment and a second biological sample is obtained after treatment (eg, surgery). A first value is obtained in the first sample, for example the Z value of the region (eg, the region methylation level and the normalized number of CNAs) and the hypomethylation and CNA (eg, amplification or deletion). It is the number of regions to be shown. The second value can be obtained in the second sample. In another embodiment, a third, or even added sample may be obtained after treatment. The number of regions exhibiting hypomethylation and CNA (eg amplification or deletion) can be obtained from the third or further added samples.

  As described above for FIGS. 40A and 41A, a first number of hypomethylated regions in the first sample can be compared to a second amount of hypomethylated regions in the second sample. . As described above for FIGS. 40B and 41B, the first amount of regions showing hypomethylation in the first sample can be compared to the second amount of regions showing hypomethylation in the second sample. . Comparison of the first amount with the second amount and comparison of the first number with the second number can be used to determine the prognosis of treatment. In various embodiments, only one of those comparisons may be a determinant of prognosis, or both comparisons may be used. In embodiments where a third or additional sample is obtained, one or more of these samples are used to determine the prognosis of the treatment, either by itself or in combination with the second sample. obtain.

  In some implementations, the prognosis is expected to be exacerbated when the first difference between the first amount and the second amount is below the first difference threshold. In another implementation, the prognosis is expected to be exacerbated when the second difference between the first number and the second number is below the second difference threshold. The thresholds may be the same or different. In one embodiment, the first difference threshold and the second difference threshold are zero. Thus, in the above example, the difference between the methylation values indicates a worse prognosis for patient TBR34.

  The prognosis may be better if the first difference and / or the second difference are above the same threshold or respective thresholds. Classification of prognosis may depend on how much the difference is above or below the threshold. Multiple thresholds can be used to give different classifications. The greater the difference, the better the prognosis can be expected, and the smaller the difference (and even negative values), the worse the prognosis can be expected.

  In some embodiments, the times at which various samples are taken are also recorded. Using such time parameters, the kinetics or rate of change of said amount can be determined. In one embodiment, a rapid decrease in tumor-associated hypomethylation in plasma and / or a rapid decrease in tumor-associated CNA in plasma is predictive of a good prognosis. Conversely, a static or rapid increase in tumor-associated hypomethylation in plasma and / or a static or rapid increase in tumor-associated CNA is predictive of a poor prognosis. Methylation and CNA measurements can be used in combination with other clinical parameters (eg imaging results or serum biochemistry or protein markers) for prediction of clinical outcome.

  In addition to plasma, other samples may be used in embodiments. For example, tumor-associated methylation abnormalities (eg, hypomethylation) and / or tumor-associated CNAs are found in circulating urine, stool, saliva, sputum, bile fluid from tumor cells in the blood of cancer patients. It can be measured from acellular DNA or tumor cells in pancreatic juice, cervical swabs, secretions from the reproductive system (eg vagina), ascites, pleural fluid, semen, sweat and tears.

  In various embodiments, tumor-associated methylation abnormalities (eg, hypomethylation) and / or tumor-associated CNAs include breast cancer, lung cancer, colorectal cancer, pancreatic cancer, ovarian cancer, nasopharyngeal cancer, cervical cancer. Cancer, melanoma, brain tumors, etc. can be detected in blood or plasma. Indeed, since methylation and genetic alterations such as CNA are universal phenomena in cancer, the approach can be used for all cancer types. Methylation and CNA measurements can be used in combination with other clinical parameters (eg, imaging results) for predicting clinical outcome. Embodiments may also be used in the screening and monitoring of patients with precancerous lesions (eg, adenomas).

  Thus, in one embodiment, the biological sample is taken prior to treatment and CNA and methylation measurements are repeated after treatment. From the measurement, a first amount after the regions determined to show deletion or amplification can be obtained, which can be obtained after the regions determined to have a region methylation level above each region cutoff value. A second amount can be obtained. The prognosis of the organism can be determined by comparing the first amount to the later first amount and the second amount to the later second amount.

  A comparison to determine the prognosis of an organism can include determining a first difference between a first amount and a subsequent first amount, where the first difference is one or more firsts. A prognosis can be determined by comparison with a difference threshold. A comparison to determine the prognosis of an organism can include determining a second difference between a second amount and a subsequent second amount, where the second difference is one or more firsts. It can be compared to two difference thresholds. The threshold can be zero or another number.

  The prognosis may be predicted to be worse when the first difference is below the first difference threshold than when the first difference is above the first difference threshold. The prognosis may be predicted to be worse when the second difference is below the second difference threshold than when the second difference is above the second difference threshold. Examples of treatments include immunotherapy, surgery, radiation therapy, chemotherapy, antibody-based therapy, gene therapy, epigenetic therapy or targeted therapy.

E. Performance The diagnostic ability of different numbers of sequence reads and different numbers of bin sizes in CNA and methylation analysis is described below.

1. Number of Sequence Reads According to one embodiment, 32 healthy control patients, 26 patients with hepatocellular carcinoma and other cancer types (eg nasopharyngeal cancer, breast cancer, lung cancer, neuroendocrine cancer and leiomyosarcoma). DNA of 20 patients with Twenty-two out of 32 healthy subjects were randomly selected as a reference group. The mean and standard deviation (SD) of these 22 reference individuals were used to determine the normal range of methylation density and genomic representation. The DNA extracted from the plasma sample of each individual was used for construction of a sequencing library using a Illumina paired-end sequencing kit. The sequencing library was then subjected to bisulfite treatment which converted unmethylated cytosine residues to uracil. The bisulfite treated sequencing library of each plasma sample was sequenced using one lane of an Illumina HiSeq2000 sequencer.

  After basecalling, adapter sequences at the ends of the fragments and a low quality base (ie, quality score less than 5) were removed. Trimmed reads in FASTQ format were then processed by a methylation data analysis pipeline called Methy-Pipe (P Jiang et al. 2010, IEEE International Conference on Bioinformatics and Biomedicine, doi: 10.1109 / BIBMW. 2010.5703866). To align the bisulfite converted sequencing reads, first, using the reference human genome (NCBI build 36 / hg19), in silico conversion of all cytosine residues to thymine separately on the Watson and Crick strands. Was done. Next, in silico conversion of each cytosine to thymine was performed in all processed reads, and positional information of each converted residue was retained. SOAP2 was used to align the converted reads to two pre-converted reference human genomes (R Li et al. 2009 Bioinformatics 25: 1966-1967), with up to 2 mismatches in each of the aligned reads. Forgiven Only reads that could be mapped to unique genomic locations were used for downstream analysis. Ambiguous and overlapping (clone) reads that map to both Watson and Crick strands were eliminated. Cytosine residues within the CpG dinucleotide sequence were used for downstream methylation analysis. After alignment, the cytosines originally present on the sequenced reads were recovered based on the positional information retained during the in silico conversion. Recovered cytosines in CpG dinucleotides were scored as methylated. Thymine in the CpG dinucleotide was scored as an unmethylated state.

  In the methylation analysis, the genome was divided into bins of equal size. The sizes of bins examined include 50 kb, 100 kb, 200 kb and 1 Mb. The methylation density of each bin was calculated as the number of methylated cytosines in the CpG dinucleotide sequence divided by the total number of cytosines at the CpG position. In other embodiments, bin sizes may be unequal across the genome. In one embodiment, each bin in such bins of unequal size is compared across multiple subjects.

Methylation densities were compared to the results of the reference group to determine if plasma methylation densities of the tested cases were normal. Twenty-two out of thirty-two healthy subjects were randomly selected as a reference group for the calculation of methylated Z-values ( _Zmeth ).

In the formula,

Is the methylation density of a particular 1 Mb bin of the tested case;

Is the mean methylation density of the corresponding bins in the reference group;

Is the SD of the methylation density of the corresponding bins in the reference group.

  In CNA analysis, the number of sequenced reads mapped to each 1 Mb bin was determined (KCA Chan el al. 2013 Clin Chem 59: 211-24). Sequencing in each bin after correction for GC bias using the Locally Weighted Scatter Plot Smoothing regression reported previously (EZ Chen et al. 2011 PLoS One 6: e21791). The lead density was determined. In plasma analysis, the sequenced read densities of the tested cases were compared to a reference group to determine the CNA Z-value,

Was calculated.

In the formula,

Is the sequenced read density of a particular 1 Mb bin of the tested case;

Is the average of the sequenced read densities of the corresponding bins of the reference group;

Was the SD of the sequenced read density of the corresponding bins in the reference group. A bin was defined as exhibiting a CNA if the Z _CNA of the bin was less than -3 or _greater than 3.

An average of 93 million aligned reads (range: 39-142 million) was obtained for each case. To evaluate the effect of reducing the number of sequenced reads on diagnostic performance, 10 million aligned reads from each case were randomly selected. The same reference population was used to establish a reference range for each 1 Mb bin in the sequenced read reduced dataset. The percentage of bins showing significant hypomethylation (ie Z _meth <-3) and the percentage of bins having CNA (ie Z _CNA <-3 or _> 3) was determined in each case. To illustrate the diagnostic power of genome-wide hypomethylation and CNA analysis on a dataset with all sequenced reads from one lane and 10 million reads per case, receiver operating characteristics (ROC) Curves were used. In the ROC analysis, all 32 healthy subjects were used in the analysis.

  FIG. 42 shows a diagram of the diagnostic capabilities of genome-wide hypomethylation analysis with different numbers of sequenced reads. In hypomethylation analysis, the area under the curve of the ROC curve is not significantly different between the two data sets analyzed for all sequenced reads from one lane and 10 million reads per case. None (P = 0.761). In CNA analysis, diagnostic power deteriorated with a significant decrease in the area under the curve when the number of sequenced reads decreased from the use of 1-lane data to 10 million (P <0. .001).

2. Impact of using different bin sizes In addition to dividing the genome into 1 Mb bins, it was also investigated whether smaller bin sizes could be used. In theory, the use of smaller bins can reduce the variability in methylation density within the bin. This is because genomic regions with different methylation densities can differ significantly. For larger bins, the probability of containing regions with different methylation densities increases, leading to an overall increase in variability in the methylation densities of the bins.

  While the use of smaller bin sizes can reduce the variability in methylation density associated with interregional differences, this, in turn, reduces the number of sequenced reads that map to a particular bin. The reduction of reads mapped to individual bins increases variability due to extraction variability. The optimal bin size that can result in the smallest overall variability in methylation density can be determined empirically, as required by the particular diagnostic application (eg, total number of sequenced reads per sample). And the type of DNA sequencer used).

  FIG. 43 shows ROC curves in cancer detection based on genome-wide hypomethylation analysis using different bin sizes (50 kb, 100 kb, 200 kb and 1 Mb). The P-values shown are P-values for area under the curve comparisons using a 1 Mb bin size. A trend of improvement can be seen when the bin size is reduced from 1 Mb to 200 kb.

F. Cumulative Probability Score The amount of regions of methylation and CNA can be various values. The above example illustrates the number of regions that exceed the cut-off value or the percentage of regions that show significant hypomethylation or CNA as a parameter for classifying whether a sample is associated with cancer. It was Such an approach does not consider the magnitude of the anomaly in individual bins. For example, a bin with Z _meth of −3.5 and a bin with Z _meth of −30 are both identical because they are both classified as having significant hypomethylation. However, the extent of hypomethylation changes in plasma (ie, the magnitude of Z _meth levels) is affected by the amount of cancer-associated DNA in the sample, thus supplementing the information on the percentage of bins showing abnormalities, It may reflect tumor burden. A higher concentration fraction of tumor DNA in plasma samples results in a lower methylation density, which translates into a lower Z _meth value.

1. Cumulative Probability Score as a Diagnostic Parameter To take advantage of the information gained from the magnitude of abnormalities, an approach called Cumulative Probability (CP) Score is developed. Based on the normal distribution probability function, each Z _meth value was converted to the probability that such an observation would be obtained by chance.

The CP score is for bin (i) with Z _meth less than -3,

Where Prob _i is the Z _meth probability of bin (i) following a Student t distribution with 3 degrees of freedom, and log is the natural logarithmic function. In another embodiment, a logarithm as low as 10 (or other number) may be used. In other embodiments, other distributions, such as, but not limited to, normal distributions and gamma distributions, can be applied to transform Z values into CP.

  Larger CP scores indicate a lower probability of having such a deviated methylation density in the normal population by chance. Therefore, a high CP score indicates that it is more likely to have aberrantly hypomethylated DNA in the sample (eg, the presence of cancer-associated DNA).

  The CP score measure has a higher dynamic range when compared to the percentage of bins that show anomalies. Tumor burden between different patients can vary widely, but a larger range of CP values is useful to reflect the tumor burden of patients with higher and lower tumor burdens. Moreover, the use of CP scores can potentially be more sensitive for detecting changes in the concentration of tumor-associated DNA in plasma. This is advantageous for monitoring treatment response and prognosis. Therefore, a decrease in CP score during treatment is indicative of a good therapeutic response. A lack of or even an increase in CP score during treatment is indicative of poor response or lack of response. In prognosis, a high CP score indicates a high tumor burden, suggesting a poor prognosis (eg, higher probability of death or tumor progression).

  FIG. 44A shows the diagnostic potential for cumulative probability (CP) and the percentage of bins with abnormalities. There was no significant difference between the areas under the curves of the two types of diagnostic algorithms (P = 0.791).

  FIG. 44B shows the diagnostic potential of plasma analysis of global hypomethylation, CG island hypermethylation and CNA. Sequencing of one lane per sample (200 kb bin size for hypomethylation analysis, 1 Mb bin size for CNA, and CG islands defined according to a database hosted by the University of California, Santa Cruz (UCSC)) The area under the curve for all three analyzes was greater than 0.90.

  In subsequent analyses, the highest CP score in control patients was used as the cutoff for each of the three analyses. Selection of these cutoffs gave 100% diagnostic specificity. The diagnostic sensitivity of global hypomethylation, CG island hypermethylation and CNA analysis was 78%, 89% and 52%, respectively. At least one of these three abnormalities was detected in 43 out of 46 cancer patients, resulting in a sensitivity of 93.4% and a specificity of 100%. These results indicate that these three analyzes can be used synergistically to detect cancer.

  FIG. 45 shows a table containing global hypomethylation, CG island hypermethylation and CNA results in patients with hepatocellular carcinoma. The CP score cutoff values in these three analyzes were 960, 2.9, and 211, respectively. Positive CP score results were bolded and underlined.

  FIG. 46 shows a table containing global hypomethylation, CG island hypermethylation and CNA results in patients with cancers other than hepatocellular carcinoma. The CP score cutoff values in these three analyzes were 960, 2.9, and 211, respectively. Positive CP score results were bolded and underlined.

2. Application of CP Score to Cancer Monitoring Serial samples were taken from HCC patient TBR34 before and after treatment. These samples were analyzed for global hypomethylation.

FIG. 47 shows a serial analysis of plasma methylation in case TBR34. The innermost ring shows the methylation density of buffy coat (black) and tumor tissue (purple). In each of these plasma samples, Z _{meth of 1} Mb bin is shown. The difference between the two lines represents a Z _meth difference of 5. Red and gray dots represent bins with hypomethylation and no change in methylation density compared to the reference group. Plasma samples taken from the second inner ring outward, pretreatment, 3 days after tumor resection and 2 months after tumor resection, respectively. Prior to treatment, a high degree of hypomethylation could be observed in plasma with> 18.5% of the bins having a Z _meth less than -10. At 3 days after tumor resection, it could be observed that the degree of hypomethylation in plasma was reduced, and neither bin had Z _meth less than -10.

  Table 5 shows that the magnitude of hypomethylation changes decreased 3 days after surgical excision of the tumor, whereas the proportion of abnormal bins showed an increase. On the other hand, the CP score may more accurately indicate the decrease in the degree of hypomethylation in plasma and better reflect the change in tumor burden.

  Even 2 months after surgery (OT), there was a significant proportion of bins showing hypomethylation changes. The CP score also remained stationary at approximately 15,000. This patient was later diagnosed with multifocal tumor deposits (previously unknown at the time of surgery) in the remaining unresected liver at 3 months and had multiple lungs 4 months after surgery. It was reported to have metastases. The patient died of metastatic disease 8 months after surgery. These results suggested that the CP score may be more powerful in reflecting tumor burden than the percentage of bins with abnormalities.

  Overall, CP may be useful in applications requiring measurement of the amount of tumor DNA in plasma. Examples of such applications include prognosis and monitoring of cancer patients (eg, to observe response to treatment or to observe tumor progression).

  The cumulative Z value is the direct sum of the Z values (ie, no conversion to probabilities). In this example, the cumulative Z value behaves the same as the CP score. In another example, CP may be more sensitive than cumulative Z-value in monitoring residual disease due to the greater dynamic range of CP scores.

X. Effect of CNAs on Methylation The use of CNAs and methylation to determine each classification of cancer level, where these classifications are combined to give a third classification, has been described above. . In addition to such combinations, CNAs are used to change the cutoff values in methylation analysis and to identify false positives by comparing the methylation levels of groups of regions with different CNA features. obtain. For example, hypermethylation level (e.g., 3 more than Z _CNA) can be compared to normal abundance of methylation levels _{(e.g., -3 <Z CNA <3)} . First, the effect of CNA on methylation levels is described.

A. Changes in Methylation Density in Regions with Chromosomal Increases and Decreases Tumor tissue generally exhibits global hypomethylation, so the presence of tumor-derived DNA in the plasma of cancer patients when compared to non-cancer subjects. , Resulting in a decrease in methylation density. The degree of hypomethylation in plasma of cancer patients is theoretically proportional to the concentration fraction of tumor-derived DNA in plasma samples.

  In the regions showing chromosomal gains in tumor tissue, additional amounts of tumor DNA are released from the amplified DNA segment into plasma. This increased contribution of tumor DNA to plasma theoretically results in a higher degree of hypomethylation in plasma DNA in the affected area. An additional factor is that genomic regions that exhibit amplification are expected to confer growth significance on tumor cells and are therefore expected to be expressed. Such regions are generally hypomethylated.

  In contrast, in regions showing chromosomal loss in tumor tissue, the reduced contribution of tumor DNA to plasma results in a lower degree of hypomethylation, as compared to regions without copy number changes. An additional factor is that the genomic region that is deleted in tumor cells may contain tumor suppressor genes, and that having such a silenced region may be beneficial to tumor cells. Therefore, it is expected that such regions are more likely to be hypermethylated.

  Here, the results of two HCC patients (TBR34 and TBR36) are used to explain this effect. Figure 48A (TBR36) and Figure 49A (TBR34) have rings highlighting regions with increased or decreased chromosomes and the corresponding methylation analysis. 48B and 49B show plots of methylated z-values for decrease, normal, and increase in TBR36 and TBR34 patients, respectively.

  FIG. 48A shows a Circos plot showing CNA (inner ring) and methylation changes (outer ring) in bisulfite-treated plasma DNA of HCC patient TBR36. Red circles highlight regions with increased or decreased chromosomes. Regions showing increased chromosomes were more hypomethylated than regions without copy number changes. Regions showing chromosomal loss had a lower degree of hypomethylation than regions without copy number changes. FIG. 48B is a plot of methylated Z-values of HCC patient TBR36 in regions with chromosomal gains and losses and in unchanged copy number. Regions with chromosomal gains have more negative Z values (lower methylation) and regions with chromosomal loss have less negative Z values (less hypomethylation) compared to regions without copy changes. Not).

  FIG. 49A shows a Circos plot showing CNA (inner ring) and methylation changes (outer ring) in bisulfite-treated plasma DNA of HCC patient TBR34. FIG. 49B is a plot of methylated Z-values of HCC patient TBR34 in regions with increased and decreased chromosomes, and in regions with unchanged copy number. The difference in methylation density between regions with chromosomal gains and losses was greater in patient TBR36 than in patient TBR34, due to the higher concentration fraction of tumor-derived DNA in patient TBR36.

  In this example, the region used to determine CNA is the same as the region used to determine methylation. In one embodiment, the respective region cutoff value depends on whether the respective regions exhibit deletions or amplifications. In some implementations, each region cutoff value (eg, the Z value cutoff used to determine hypomethylation) is greater when each region shows amplification than when no amplification is shown. Have a scale (eg, scale can be greater than 3 and cutoffs less than −3 can be used). Thus, in the hypomethylation test, each region cutoff value may have a greater negative value when each region shows amplification than when no amplification is shown. Such practices are expected to improve the specificity of tests for detecting cancer.

  In another implementation, each region cutoff value has a smaller magnitude (eg, less than 3) when each region exhibits a deletion than when no deletion is exhibited. Thus, in the hypomethylation test, each region cutoff value may have a less negative value when each region shows a deletion than when no deletion is shown. Such practices are expected to improve the sensitivity of tests for detecting cancer. The adjustment of the cutoff value in the above implementation may vary depending on the desired sensitivity and specificity of the particular diagnostic scenario. In other embodiments, methylation and CNA measurements can be used in combination with other clinical parameters (eg, imaging results or serum biochemistry) to predict cancer.

B. Use of CNAs to Select Regions As noted above, plasma methylation densities were shown to be altered in regions with copy number abnormalities in tumor tissue. Increased contribution of hypomethylated tumor DNA to plasma in regions with increased copy number in tumor tissue results in a greater degree of hypomethylation of plasma DNA compared to regions without copy number abnormalities . Conversely, in regions with copy number reduction in tumor tissue, the reduced contribution of hypomethylated cancer-derived DNA to plasma results in lesser hypomethylation of plasma DNA. This association between the methylation density and relative presentation of plasma DNA is potentially a consequence of hypomethylation associated with the presence of cancer-associated DNA and other non-cancerous hypomethylation in plasma DNA. Can be used to distinguish from sexual causes (eg, SLE).

  To illustrate this approach, plasma samples from two hepatocellular carcinoma (HCC) patients and two patients without cancer but with SLE were analyzed. These two SLE patients (SLE04 and SLE10) showed hypomethylation in plasma and the apparent presence of CNA. In patient SLE04, 84% of the bins had hypomethylation and 11.2% had CNA. In patient SLE10, 10.3% of the bins had hypomethylation and 5.7% had CNA.

50A and 50B show the results of plasma hypomethylation and CNA analysis for SLE patients SLE04 and SLE10. The outer ring shows the methylated Z value (Z _meth ) at 1 Mb resolution. Bins with methylation Z _meth less than -3 is red, the bottle having a -3 greater than Z _meth was gray. The inner ring shows the CNA Z value (Z _CNA ). Green, red and gray dots represent bins with greater than 3, less than 3, and -3 to 3 Z _CNAs , respectively. Hypomethylation and CNA changes were observed in plasma in these two SLE patients.

To determine whether changes in methylation and CNA are consistent with the presence of cancer-derived DNA in plasma, Z _{meth in} regions with more than 3, less than 3, and -3 to 3 Z _CNAs were compared. It was In the methylation changes and CNAs contributed by cancer-derived DNA in plasma, regions with Z _CNAs less than -3 were predicted to be less hypomethylated, with less negative Z _meth I had. In contrast, regions with Z _CNAs greater than 3 were expected to be more hypomethylated and had more negative Z _meth . For purposes of illustration, a one-sided rank sum test is applied such that Z _{meth of} regions with _CNA (ie, regions with Z _CNA less than −3 or greater than 3) has no CNA (ie, −3). Z _{meth of the} region having a Z _CNA of ˜3). In other embodiments, other statistical tests may be used, including but not limited to Student's t test, analysis of variance (ANOVA) test and Kruskal-Wallis test.

51A and 51B show Z _meth analyzes for the regions with and without CNA in the plasma of two HCC patients (TBR34 and TBR36). Region having an area and 3 greater than Z _CNA having Z _CNA less than -3, respectively, represent the region with the plasma, the region and overrepresentation having under-presentation. In both TBR34 and TBR36, plasma with under-presented area (i.e., a region having a Z _CNA less than -3), the region having a normal presentation in plasma (i.e., the Z _CNA of -3 to 3 _Had a significantly higher Z _meth (P value <10 ⁻⁵ , one-sided rank sum test). Normal presentation is consistent with that expected for the euploid genome. In regions with over-presentation in plasma (ie regions with Z _CNA greater than 3), they had significantly lower Z _meth than in regions with normal presentation in plasma (P value <10 ^−5). , One-sided rank sum test). All these changes were consistent with the presence of hypomethylated tumor DNA in plasma samples.

51C and 51D show Z _meth analyzes for the regions with and without CNA in the plasma of two SLE patients (SLE04 and SLE10). Region having an area and 3 greater than Z _CNA having Z _CNA less than -3, respectively, represent the region with the plasma, the region and overrepresentation having under-presentation. In SLE04, plasma with under-presented area (i.e., a region having a Z _CNA less than -3), from regions with normal presentation in plasma (i.e., a region having a Z _CNA of -3 to 3) Also had no significantly higher Z _meth (P value = 0.99, one-sided rank sum test) and regions with over-presentation in plasma (ie regions with Z _CNA > 3) were normal in plasma. We did not have a significantly lower Z _meth than the region with presentation (P value = 0.68, one-sided rank sum test). These results differed from the expected changes due to the presence of tumor-derived hypomethylated DNA in plasma. Similarly, in SLE10, regions having a Z _CNA less than -3, than the region having the Z _CNA of -3 to 3, did not have a high Z _meth significantly (P value = 0.99, one-sided rank Sum test).

The reason for the lack of a typical cancer-related pattern between Z _meth and Z _CNA in SLE patients is the absence of CNAs in certain cell types that also show hypomethylation in SLE patients. Indeed, the observed presence of CNA and hypomethylation is due to the altered size distribution of circulating DNA in SLE patients. The change in size distribution can potentially change the sequenced read densities in different genomic regions, resulting in an apparent CNA, as the reference is obtained from healthy subjects. As described in the previous section, there is a correlation between the size of the circulating DNA fragment and its methylation density. Thus, changes in size distribution can also result in aberrant methylation.

Although areas with 3 more than Z _CNA had a slightly lower methylation level than the region having the Z _CNA of -3 to 3, p values in comparison, than that observed in two cancer patients Was also much higher. In one embodiment, the p-value can be used as a parameter to determine the likelihood that a tested case has cancer. In another embodiment, the difference in Z _meth between regions with normal and abnormal presentation can be used as a parameter to indicate the potential presence of cancer. In one embodiment, a panel of cancer patients is used to establish a correlation between Z _meth and Z _CNA , as well as to determine thresholds for different parameters, wherein the changes are derived from cancer in a test plasma sample. It can be shown to be consistent with the presence of hypomethylated DNA.

Thus, in one embodiment, CNA analysis can be performed to determine a first series of regions, all of which exhibit one of deletion, amplification, or normal presentation. For example, the first series of regions may exhibit all deletions, or all amplifications, or all may exhibit normal presentation (eg, have a normal first amount of regions (eg, Normal Z _meth )). A methylation level can be determined for this first series of regions (eg, the first methylation level of method 2800 can correspond to the first series of regions).

  CNA analysis can determine a second series of regions, all showing a second deletion, amplification, or normal presentation. The second set of regions is shown as different from the first set of regions. For example, if the first set of regions was normal, the second set of regions may exhibit deletions or amplifications. The second level of methylation can be calculated based on each of the number of methylated DNA molecules at a site within the second series of regions.

  Parameters can then be calculated between the first methylation level and the second methylation. For example, the difference or ratio can be calculated and compared to the cutoff value. Also, the difference or ratio can be multiplied by a probability distribution (eg, as part of a statistical test) to determine the probability of obtaining that value, which is compared to the cutoff value to determine the methylation level. The level of cancer can be determined based on the. Such cutoffs can be selected to distinguish between samples with cancer and samples without cancer (eg, SLE).

  In one embodiment, the methylation level of a first series of regions or a mixture of regions (ie, a mixture of regions showing amplification, deletion, and normal) can be determined. This methylation level can then be compared to the first cutoff as part of the first step of the analysis. Above the cutoff, this indicates a cancer potential, but the above analysis can be performed to determine if the manifestation is a false positive. Thus, the final classification of cancer levels may include comparison of the two methylation level parameters with a second cutoff.

  The first methylation level can be a statistic (eg, mean or median) of the calculated region methylation levels for each region of the first series of regions. The second methylation level can also be a statistic of the calculated region methylation level for each region of the second series of regions. As an example, the statistic may be determined using a one-tailed rank sum test, Student's t-test, analysis of variance (ANOVA) test, or Kruskal-Wallis test.

XI. Cancer Type In addition to determining whether an organism has cancer, in embodiments, the cancer type associated with the sample can be identified. Global hypomethylation, CG island hypermethylation, and / or CNA patterns can be used in identifying this cancer type. The pattern may include clustering of patients with a known diagnosis using the measured regional methylation levels, the respective CNA values of the regions, and the CG island methylation levels. The results below show that organisms with similar cancer types have similar values in the region and CG islands, and non-cancer patients have similar values. In clustering, each of these values in a region or island can be a separate dimension in the clustering process.

  The same cancer types are known to share similar genetic and epigenetic changes (E Gebhart et al. 2004 Cytogenet Genome Res; 104: 352-358; PA Jones et al. 2007 Cell; 128: 683-692). Below is described how the patterns of CNA and methylation changes detected and detected in plasma are useful in inferring the origin or type of cancer. Plasma DNA samples from HCC patients, non-HCC patients and healthy control patients were classified, for example using hierarchical clustering analysis. The analysis is performed by, for example, the heatmap. It was performed using two functions (cran.r-project.org/web/packages/gplots/gplots.pdf).

  To illustrate the potential of this approach, two sets of criteria (Group A and Group B) were used as an example to identify features useful for classifying plasma samples (see Table 6). ). In other embodiments, other criteria may be used to identify these features. Features used included global CNA at 1 Mb resolution, global methylation density at 1 Mb resolution, and CG island methylation.

  In the first two examples, CNA, global methylation at 1 Mb resolution and CG island methylation features were all used for classification. In other embodiments, other criteria may be used, such as, but not limited to, the accuracy of measuring characteristics in the plasma of the reference group.

  FIG. 52A shows HCC patients, non-HCC cancer patients and healthy subjects, using all of the CNAs of 355, global methylation characteristics at 1 Mb resolution of 584 and group A characteristics of 1,130 including the methylation status of 110 CG islands. 3 shows a hierarchical clustering analysis for plasma samples obtained from control patients. The upper color bars represent sample groups: green, blue and red represent healthy subjects, HCC cancer patients and non-HCC cancer patients, respectively. Overall, the three subject groups tended to cluster together. The vertical axis represents classification characteristics. Features with similar patterns among different subjects were clustered together. These results indicate that CG island methylation changes in plasma, genome-wide methylation changes at 1 Mb resolution, and patterns of CNAs could be used to identify the origin of cancer in potentially unknown primary patients. Suggests that.

  Figure 52B shows HCC patients, non-HCC cancer patients and all 2,780 Group B features, including 759 CNAs, 1,911 global methylation at 1 Mb resolution and 191 CG island methylation status. 3 shows a hierarchical clustering analysis for plasma samples obtained from healthy control patients. The upper color bars represent sample groups: green, blue and red represent healthy subjects, HCC cancer patients and non-HCC cancer patients, respectively. In general, the three subject groups tended to cluster together. The vertical axis represents classification characteristics. Features with similar patterns among different subjects were clustered together. These results show that patterns of CG island methylation changes in plasma, genome-wide methylation changes at 1 Mb resolution, and various assembly of CNAs can be used to identify the origin of cancer in patients with unknown primary. Suggests that. The selection of classification features can be tailored for a particular application. In addition, cancer type predictions may be weighted depending on the prior probabilities of the subject for different cancer types. For example, patients with chronic viral hepatitis tend to develop hepatocellular carcinoma and addictive smokers tend to develop lung cancer. Thus, the cancer type weighted probabilities can be calculated using, for example, without limitation, logistic regression, multiple regression, or clustering regression.

  In other embodiments, one type of feature may be used for classification analysis. For example, in the following example, only global methylation at 1 Mb resolution, only CG island hypermethylation or only CNA at 1 Mb resolution is used for hierarchical clustering analysis. The distinguishing ability can be different if different features are used. Feature refinement of classification features can potentially improve classification accuracy.

  Figure 53A shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients using the Group A CG island methylation feature. In general, cancer patients were clustered together and included in non-cancer subjects. However, HCC and non-HCC patients were less segregated compared to using all three characteristics.

  FIG. 53B shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients using group A global methylation density at 1 Mb resolution as the classification feature. A preferential clustering of HCC and non-HCC patients was observed.

  FIG. 54A shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using Group A whole CNAs at 1 Mb resolution as the classification feature. A preferential clustering of HCC and non-HCC patients was observed.

  FIG. 54B shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients using Group B CG island methylation density as a classification feature. A preferential clustering of HCC and non-HCC cancer patients could be observed.

  FIG. 55A shows a hierarchical clustering analysis for plasma samples from HCC patients, non-HCC cancer patients and healthy control patients, using global methylation density at Group 1 Mb resolution as a classification feature. A preferential clustering of HCC and non-HCC cancer patients could be observed.

  FIG. 55B shows a hierarchical clustering analysis for plasma samples obtained from HCC patients, non-HCC cancer patients and healthy control patients using global CNA at Group B 1 Mb resolution as the classification feature. A preferential clustering of HCC and non-HCC cancer patients could be observed.

  The results of these hierarchical clusterings on plasma samples suggest that different feature combinations could potentially be used for primary cancer type identification. The accuracy of classification can be further improved by further subdividing the selection criterion.

  Thus, in one embodiment, if the methylation classification indicates that a cancer is present in the organism, the cancer type associated with the organism is the methylation level (eg, the first methylation or any region from method 2800). Methylation level) can be determined by comparing the corresponding values determined from other organisms (ie other organisms of the same species, such as humans). The corresponding value can be the value of the same region or series of sites where the methylation level was calculated. At least two other organisms are identified as having different cancer types. For example, the corresponding values can be organized into clusters, where the two clusters are associated with different cancers.

  Furthermore, when CNA and methylation are used together to obtain a third classification of cancer levels, the CNA and methylation characteristics can be compared to corresponding values obtained from other organisms. For example, a first amount of regions exhibiting deletions or amplifications (eg, from Figure 36) is compared to corresponding values determined from other organisms to identify cancer types associated with that organism. obtain.

  In some embodiments, the methylation characteristic is regional methylation level of multiple regions of the genome. Regions that have been determined to have regional methylation levels that exceed their respective regional cutoff values can be used, for example, the regional methylation level of an organism is the regional methylation of another organism in the same region of the genome. Can be compared to the level. The comparison may allow cancer type discrimination, or may provide an additional filter to identify cancer (eg, to identify false positives). Thus, whether the organism has the first cancer type, the absence of cancer, or the second cancer type can be determined based on the comparison.

  Other organisms (along with the organism under test) can be clustered using regional methylation levels. Therefore, comparison of regional methylation levels can be used to determine which cluster an organism belongs to. Clustering may also use the CNA normalized number of regions determined to exhibit deletions or amplifications, as described above. The clustering may then use the methylation density of each of the hypermethylated CG islands.

  To illustrate the principle of this method, an example using logistic regression for the classification of two unknown samples is shown. The purpose of this classification was to determine if these two samples were HCC or non-HCC cancer. A training sample set was collected that included 23 plasma samples taken from HCC patients and 18 samples taken from patients with cancers other than HCC. Therefore, there were a total of 41 cases in the training set. In this example, five features for CG island methylation (X1-X5), six features for 1 Mb region methylation (X6-X11) and two features for 1 Mb region CNA (X12-X13) are included, Thirteen features were selected. CpG methylation features were selected based on the criteria of at least 15 cases in the training set with Z values greater than 3 or less than -3. A 1 Mb methylation feature was selected based on the criteria of at least 39 cases in the training set with Z values greater than 3 or less than -3. CNA features were selected based on criteria for at least 20 cases with Z values greater than 3 or less than -3. Logistic regression was performed on the samples in this training set to determine the regression coefficient for each feature (X1-X13). Features with regression coefficients of greater magnitude (whether positive or negative) give better discrimination between HCC and non-HCC samples. The Z value for each case in each feature was used as the input value for the independent variable. Next, two plasma samples, one from an HCC patient (TBR36) and the other from a patient with lung cancer (TBR177), were analyzed for 13 features.

  In this cancer typing analysis, it was hypothesized that these two samples were taken from patients with cancers of unknown primary. In each sample, the Z value of each feature is input to a logistic regression equation to determine the natural logarithm of the odds ratio (ln (odds ratio)), where the odds ratio has the probability of having HCC and HCC. Represents the ratio of no probabilities (HCC / non-HCC).

  Table 7 shows the regression coefficients for the 13 features of the logistic regression equation. The Z values for each feature of the two examined cases (TBR36 and TBR177) are also shown. The HCC ln (odds ratio) of TBR36 and TBR177 were 37.03 and -4.37, respectively. From these odds ratios, the likelihood that plasma samples were taken from HCC patients was calculated to be> 99.9% and 1%, respectively. In summary, TBR36 was likely to be a sample taken from HCC patients, while TBR177 was unlikely to be a sample taken from HCC patients.

  In other embodiments, hierarchical clustering regression, classification tree analysis and other regression models can be used to determine potential cancer primary sites.

XII. Materials and Methods A. Preparation and sequencing of bisulfite-treated DNA library Genomic DNA (5 μg) supplemented with 0.5% (w / w) unmethylated λDNA (Promega) was used as Covaris S220 System (Covaris). Fragmented to approximately 200 bp. A DNA library was prepared using the Paired-End Sequencing Sample Preparation Kit (Illumina) according to the manufacturer's instruction manual, except that the methylated adapter (Illumina) was ligated to the DNA fragment. After two rounds of purification with AMPure XP magnetic beads (Beckman Coulter), the ligation product was split into two parts, one of which was twice bisulfite modified with EpiTect Bisulfite Kit (Qiagen). I went to Unmethylated cytosine at the CpG site in the insert was converted to uracil, leaving methylated cytosine intact. Adapter-ligated DNA molecules treated with or without sodium bisulfite were enriched in 10 cycles of PCR using the following recipe: 2.5 U of PfuTurboCx hot start DNA polymerase (50 μl reaction). (Agilent Technology), 1 × PfuTurboCx reaction buffer, 25 μM dNTP, 1 μl PCR Primer PE 1.0 and 1 μl PCR Primer PE 2.0 (Illumina). The thermal cycle profile was 95 ° C for 2 minutes, 98 ° C for 30 seconds, then 98 ° C for 15 seconds, 60 ° C for 30 seconds and 72 ° C for 4 minutes for 10 cycles, and the final step 72 ° C for 10 minutes. (R Lister, et al. 2009 Nature; 462: 315-322). PCR products were purified using AMPure XP magnetic beads.

  Plasma DNA extracted from 3.2-4 ml maternal plasma samples was spiked with fragmented lambda DNA (25 pg / ml plasma) and subjected to library construction as described above (RWK Chiu et al. 2011 BMJ; 342: c7401). After ligation to the methylated adapter, the ligation product was split into two equal parts and one part was subjected to two bisulfite modifications. The bisulfite treated or untreated ligation product was then enriched as above with 10 cycles of PCR.

  The bisulfite treated or untreated DNA library was sequenced on a HiSeq 2000 instrument (Illumina) in paired-end format for 75 bp. A DNA cluster was prepared using a Paired-End Cluster Generation Kit v3 on a cBot device (Illumina). Real-time image analysis and base calling were performed using HiSeq Control Software (HCS) v1.4 and Real Time Analysis (RTA) Software v1.13 (Illumina), whereby the automatic matrix and phase calculation performed on the DNA library. Based on PhiX control v3, spiked in, sequenced using.

B. After sequence alignment and methylated cytosine specific base call, the fragment end adapter sequences and low quality bases (ie, quality score <20) were removed. Trimmed reads in FASTQ format were then processed by a methylation data analysis pipeline called Methy-Pipe (IEEE International Conference on Bioinformatics and Biomedicine Workshops, Hong Kong, December 18-21, 2010). P Jiang, et al. Methy-Pipe: An integrated bioinformatics data analysis pipeline for whole genome methylome analysis). To align the bisulfite-converted sequencing reads, first in silico conversion of all cytosine residues to thymine separately on the Watson and Crick strands using the reference human genome (NCBI build 36 / hg18). Was done. Next, in silico conversion of each cytosine to thymine was performed in all processed reads, and positional information of each converted residue was retained. SOAP2R Li, et al. 2009 Bioinformatics; 25: 1966-1967) was used to align the converted reads to two pre-converted reference human genomes, with up to two mismatches in each of the aligned reads. Forgiven Only reads that could be mapped to unique genomic locations were selected. Ambiguous reads that map to both the Watson and Crick strands as well as duplicate (clone) reads with identical start and stop genomic positions were excluded. Sequenced reads with insert sizes of 600 bp or less were retained for methylation and size analysis.

  Cytosine residues within the CpG dinucleotide sequence have been the main target for downstream DNA methylation studies. After alignment, the cytosines originally present on the sequenced reads were recovered based on the positional information retained during the in silico conversion. Recovered cytosines in CpG dinucleotides were scored as methylated. Thymine in the CpG dinucleotide was scored as an unmethylated state. The unmethylated λ DNA included during library construction served as an internal standard to estimate the efficiency of sodium bisulfite modification. If the bisulfite conversion efficiency is 100%, then all the cytosines on λDNA should be converted to thymine.

XIII. Overview The use of the embodiments described herein allows for non-invasive screening, detection, monitoring or prognosis of cancer, eg, with plasma of a subject. Estimating the methylation properties of fetal DNA obtained from maternal plasma also allows prenatal screening, diagnosis, investigation or monitoring of the fetus to be performed. To illustrate the power of this approach, it was shown that the information previously obtained through placental tissue studies can be evaluated directly from maternal plasma. For example, imprinted status of loci, identification of loci with differential methylation between fetal and maternal DNA, and gestational variation in the methylation characteristics of loci have been demonstrated through direct analysis of maternal plasma DNA. To be achieved. The main advantage of this approach is that the fetal methylome can be comprehensively evaluated during pregnancy without the need for confusion with pregnancy or invasive sampling of fetal tissue. Given the changes in DNA methylation status and the known associations between many pregnancy-related conditions, the approach described in this study was designed to investigate the pathophysiology of biomarkers in those conditions and identify those biomarkers. Can serve as an important means for By focusing on imprinted loci, it was shown that both paternally transmitted fetal and maternally transmitted fetal methylation properties can be evaluated from maternal plasma. This approach could potentially be useful in the study of imprinted disease. Embodiments can also be directly applied to prenatal assessment of fetal or pregnancy related disorders.

  It has been shown that genome-wide bisulfite sequencing can be applied to study the DNA methylation properties of placental tissues. There are approximately 28 M CpG sites in the human genome (C Clark et al. 2012 PLoS One; 7: e50233). Bisulfite sequencing data for CVS and late pregnancy placental tissue specimens covered more than 80% of CpG. This represents a substantially wider coverage than can be achieved with other high throughput platforms. For example, the Illumina Infinium HumanMethylation 27K bead chip array used in the previous study on placental tissue (T Chu et al. 2011 PLoS One; 6: e14723) covers 0.1% of CpG in the genome. It was only done. More recently available the Illumina Infinium HumanMethylation 450K bead chip array only coated 1.7% of CpG (C Clark et al. 2012 PLoS One; 7: e50233). Since the MPS approach has no restrictions on probe design, hybridization efficiency or strength of antibody capture, it was possible to evaluate CpG in and out of CG islands and within most sequence configurations.

XIV. Computer System Any of the computer systems mentioned herein may utilize any suitable number of subsystems. An example of such a subsystem is shown in FIG. 33 at computing device 3300. In some embodiments, a computer system may include a single computer device, subsystems of which may be components. In other embodiments, a computer system may include multiple computer devices, each subsystem having internal components.

  The subsystems shown in FIG. 33 are interconnected via a system bus 3375. Additional subsystems such as a printer 3374, a keyboard 3378, a storage device 3379, a monitor 3376 coupled with a display adapter 3382 are shown. Peripheral devices and input / output (I / O) devices coupled to I / O control device 3371 may be coupled to the computer system by any number of means known in the art, such as serial port 3377. For example, serial port 3377 or external interface 3381 (eg, Ethernet, Wi-Fi, etc.) may be used to couple computer system 3300 to a wide area network such as the Internet, a mouse input device, or a scanner. Interconnects via the system bus 3375 allow the central processing device 3373 to communicate with each subsystem, execute instructions from the system memory 3372 or storage device 3379 (eg, fixed disk), and exchange information between subsystems. Allows you to control. System memory 3372 and / or storage device 3379 may include computer-readable media. Any of the values described herein can be output from one component to another and can be output to the user.

  A computer system may include multiple identical components or subsystems interconnected by, for example, an external interface 3381 or an internal interface. In some embodiments, computer systems, subsystems, or devices may communicate over a network. In such a device, one computer may be considered a client and another computer may be considered a server, where each may be part of the same computer system. The client and server may each include multiple systems, subsystems, or components.

  Any of the embodiments of the invention include hardware (eg, an application specific integrated circuit or field programmable gate array) and / or computer software having a generally programmable processor in a modular or integrated fashion. It will be appreciated that the implementation can be used in the form of control logic. As used herein, a processor includes a multi-core processor or the same integrated chip, or multiple processing units on a single circuit board or network connection board. Based on the disclosure and teachings provided herein, those of ordinary skill in the art will be aware of other ways and / or methods for implementing embodiments of the invention using hardware and combinations of hardware and software. You will understand.

  Any of the software components or functionality described herein may be implemented by a processor, eg, using conventional or object oriented techniques, eg, any suitable computer language such as Java, C ++ or Perl. It can be implemented as coded software. Coded software may be stored as a series of instructions or instructions on a computer-readable medium for storage and / or transmission, suitable medium including readable and writable memory (RAM), read only memory (ROM), hard disk. A magnetic medium such as a drive or a floppy disk, an optical medium such as a compact disk (CD) or a DVD (digital versatile disk), a flash memory and the like are included. Computer readable media can be any combination of such storage or communication devices.

  Such programs may also be coded and communicated using carrier signals suitable for transmission over wired, optical, and / or wireless networks compatible with various protocols, including the Internet. Therefore, a computer-readable medium according to an embodiment of the present invention can be made using a data signal encoded by such a program. Computer readable media encoded with the encoded program can be packaged with a compatible device or provided separately from other devices (eg, via Internet download). Any such computer-readable medium may reside on or within a single computer program product (eg, a hard drive, a CD, or an entire computer system), on a different computer program product in a system or network, or in that computer program product. Can be internal. The computer system can include a monitor, printer, or other suitable display for providing a user with any of the results described herein.

  Any of the methods described herein can be fully or partially performed with a computer system that includes one or more processors that can be configured to perform those steps. Accordingly, embodiments may relate to computer systems configured to perform any step of the methods described herein, which may include different components performing each step or groups of steps. Although provided as numbered steps, the method steps herein may be performed simultaneously or in different orders. Moreover, some of these steps may be used in conjunction with some of the other steps from other methods. Also, all or some of the steps may be optional. Moreover, any step of any method may be performed using modules, circuits, or other means for performing these steps.

  The specific details of the particular embodiments may be combined in any suitable manner without departing from the spirit and scope of the embodiments of the invention. However, other embodiments of the invention may relate to particular embodiments relating to particular aspects, or particular combinations of these individual aspects.

  The foregoing description of the exemplary embodiments of the present invention has been presented for purposes of illustration and explanation. It is not intended to be exhaustive or to limit the invention to the precise form recited, and many modifications and variations are possible in light of the above teaching. The embodiments best explain the principles of the invention and its practical application so that others skilled in the art may make various modifications in various embodiments and appropriate for the particular application contemplated. It has been chosen and described in order to best utilize the invention.

  References to "a", "an" or "the" are intended to mean "one or more" unless stated otherwise.

  All patents, patent applications, publications, and statements mentioned herein are incorporated by reference in their entirety for all purposes. None of them admits to be prior art.

Claims (43)

A method for determining a cancer type by analyzing a cell-free DNA molecule derived from a biological sample of an organism,
(A) obtaining a sequence read from a cell-free DNA molecule from a biological sample, wherein the sequence read comprises methylation status at one or more sites of each of the cell-free DNA molecules;
(B ) analyzing the sequence reads to determine the number of methylated cell-free DNA molecules at each of the multiple sites and obtaining the methylation status of the multiple sites in the cell-free DNA molecule by single nucleotide analysis Using the number to :
(C) determining a methylation property from methylation states of a plurality of sites; and (d) determining a cancer type based at least in part on the methylation property , wherein the cancer type is determined. Comprising comparing the methylation profile to one or more reference methylation profiles, wherein at least one of the one or more reference methylation profiles is at least 1 derived from a subject known to have cancer. Obtained from one sample,
Including the method.   The method of claim 1, wherein the biological sample is selected from the group consisting of blood, plasma, and serum.   The biological sample is a group consisting of urine, vaginal fluid, uterus or flushing fluid, multiple body fluids, ascites, cerebrospinal fluid, saliva, sweat, tears, sputum, bronchoalveolar lavage fluid, body fluid, and feces. The method of claim 1, selected from: The method according to claim 1, wherein the cancer type is based on an organ in which cancer has developed. 4. The method of any one of claims 1-3, wherein the cancer type is based on the tissue of origin of the cancer.   The tissue of cancer origin is selected from the group consisting of brain, bone, lung, heart, kidney, liver, breast, colorectal, prostate, nasopharynx, stomach, testis, skin, ovary, pancreas, uterus, and bladder. The method according to claim 5, wherein   Cancer types include lung cancer, breast cancer, colorectal cancer, prostate cancer, nasopharyngeal cancer, gastric cancer, testicular cancer, skin cancer, cancer that develops in nervous system, bone cancer, ovarian cancer, liver The method according to any one of claims 1 to 6, which is selected from the group consisting of cancer, hematopoietic tumor, pancreatic cancer, endometrial cancer, and renal cancer.   8. The method of any one of claims 1-7, further comprising sequencing the cell-free DNA molecule to obtain a sequence read.   9. The method of claim 8, wherein the sequencing comprises methylation recognition sequencing.   The method of claim 9, wherein methylation recognition sequencing comprises bisulfite sequencing.   11. The method of claim 10, wherein the bisulfite sequencing comprises genome-wide bisulfite sequencing. 12. The method of any one of claims 9-11, wherein the methylation recognition sequencing comprises paired -end massively parallel bisulfite sequencing or whole genome sequencing.   13. The method of any one of claims 8-12, further comprising enrichment for cell-free DNA molecules prior to sequencing.   14. The method of claim 13, wherein the enrichment comprises the use of hybridization probes, polymerase chain reaction amplification, or solid phase hybridization.   15. The method of any one of claims 1-14, wherein the plurality of sites comprises one or more CpG sites.   16. The method of any one of claims 1-15, wherein the plurality of sites comprises one or more CpG sites organized into one or more CpG islands.   Determining the methylation profile of multiple sites comprises calculating the methylation density of one or more CpG islands using the methylation status determined from sequence reads at one or more CpG islands. Item 16. The method according to Item 16.   Determining the methylation profile of the plurality of sites comprises calculating the methylation density of each of the plurality of sites using the methylation status determined from the sequence reads at the site. 16. The method according to 16.   19. The method of any one of claims 1-18, wherein the analysis further comprises determining the location of the cell-free DNA molecule in the genome.   Determining methylation properties determines one or more methylation levels in one or more genomic regions using methylation status determined from sequence reads of cell-free DNA molecules located in one or more genomic regions 20. The method of claim 19, comprising:   21. The method of claim 20, wherein the one or more genomic regions comprises a plurality of genomic bins identified in the genome.   22. The method of claim 21, wherein the multiple genomic bins have a size of 50 kb to 1 Mb.   23. The method of claim 21 or 22, wherein the plurality of genomic bins have different sizes. 24. Any of claims 21-23, further comprising selecting a group of genomic bins by comparing the methylation level of each of the genomic bins with a cut-off value indicating that the genomic bin is hypomethylated or hypermethylated. The method described in paragraph 1.   25. The method of claim 24, further comprising, at least in part, determining the cancer type based on the methylation level of the selected population of genomic bins. Other subjects, Ru target der known to have a cancer, the method according to claim 25. Other subject Ru healthy subject der A method according to claim 25. 28. The method of any one of claims 1-27, wherein the one or more reference methylation properties are obtained from one or more reference samples containing cell-free DNA molecules . 29. The method of claim 28, wherein another of the one or more reference methylation properties corresponds to a healthy subject. One or more reference samples includes multiple reference samples corresponding to each subject, and comparing the methylation profile to one or more reference methylation profiles is
Clustering reference methylation profiles determined from multiple reference samples to obtain one or more clusters of reference methylation profiles; and
Comparing methylation profiles with one or more categories to determine cancer types associated with methylation profiles
29. The method of claim 28, comprising :   31. The method of any one of claims 1-30, further comprising determining the presence or absence of copy number abnormalities based on methylation characteristics.   32. The method of any one of claims 1-31, further comprising determining one or more methylation levels at one or more genomic sites in a cell-free DNA molecule.   33. The method of any one of claims 1-32, wherein determining the methylation profile comprises determining one or more methylation levels of one or more CpG islands. A method for analyzing a biological sample derived from a subject, comprising:
(A) obtaining a sequence read of a cell-free DNA molecule from a biological sample of interest, wherein the sequence read comprises methylation status at multiple sites of the cell-free DNA molecule in a single nucleotide analysis;
(B) analyzing sequence reads to determine the number of cell-free DNA molecules methylated at each of the multiple sites, and determining the methylation properties of the multiple sites based on the methylation status of the multiple sites And (c) determining a tissue source of at least a portion of the cell-free DNA molecules from the biological sample based on the methylation characteristics , wherein the tissue source is Determining includes comparing the methylation profile to one or more reference methylation profiles, at least one of the one or more reference methylation profiles was obtained from a subject known to have cancer. Obtained from at least one sample .   35. The method of claim 34, wherein at least a portion of the cell-free DNA molecule is a cancer-derived molecule and at least a portion further comprises determining the cancer type of interest based on the methylation profile.   Lung cancer, breast cancer, colorectal cancer, prostate cancer, nasopharyngeal cancer, gastric cancer, testicular cancer, skin cancer, cancer that develops in nervous system, bone cancer, ovarian cancer, liver 36. The method of claim 35, selected from the group consisting of cancer, hematopoietic tumor, pancreatic cancer, endometrial cancer, and renal cancer.   35. The method of claim 34, further comprising sequencing the cell-free DNA molecule to obtain a sequence read.   38. The method of claim 37, wherein the sequencing comprises methylation recognition sequencing.   39. The method of claim 38, wherein methylation recognition sequencing comprises bisulfite sequencing.   40. The method of claim 39, wherein the plurality of sites comprises one or more CpG sites.   A program that causes a computer to execute the method according to any one of claims 1 to 7, 15 to 36, and 40.   An apparatus comprising one or more processors and a computer-readable medium storing a program according to claim 41.   A computer device comprising one or more processors programmed to perform the method of any one of claims 1-7, 15-36 and 40.